@ucsdcme @UCSanDiegoCME ucsandiegocme UC San Diego CME COVID-19 Dynamics & Evolution October 19 - October 20, 2020

cpd.ucsd.edu/covid19 #COVIDdynamics

@ucsdcme @UCSanDiegoCME ucsandiegocme UC San Diego CME AGENDA

Monday, October 19, 2020

Session 1 - Epidemiology of COVID-19 Session Chairs: Ben Radar & Michael Levy

6:00 AM Welcome & Introductions

6:10 AM THE ROLE OF BEHAVIOR, MOBILITY, AND SOCIAL NETWORKS IN SHAPING THE COVID- 19 PANDEMIC, *Sam Scarpino

6:30 AM LOCALIZED END-OF-OUTBREAK DETERMINATION FOR COVID-19: EXAMPLES FROM CLUSTERS IN JAPAN, *Natalie Linton

6:45 AM IMPROVED ESTIMATION OF TIME-VARYING REPRODUCTION NUMBERS AT LOW CASE INCIDENCE AND BETWEEN EPIDEMIC WAVES, *Kris Parag

7:00 AM PRE-SYMPTOMATIC TRANSMISSION OF SARS-COV-2 IN CHINA: BEFORE AND AFTER THE LOCKDOWN, *Mary Bushman

7:15 AM GENOMIC EPIDEMIOLOGY APPROACH TO ESTIMATING THE SERIAL INTERVAL DISTRIBUTION, *Kurnia Susvitasari

7:30 AM REVEALING THE EXTENT OF THE COVID-19 PANDEMIC IN KENYA BASED ON SEROLOGICAL AND PCR-TEST DATA, *Samuel Brand

7:45 AM ESTIMATING COVID-19 SECONDARY ATTACK RISK IN ELDERLY CARE FACILITIES IN FRANCE, *Bastien Reyné

8:00 AM ESTIMATING CUMULATIVE INCIDENCE OF SARS-COV-2 WITH IMPERFECT SEROLOGICAL TESTS: EXPLOITING CUTOFF-FREE APPROACHES, *Judith Bouman

8:15 AM BREAK

Session 2 - Characterizing SARS-CoV2-Evolution Session Chairs: Sarah Nadeau & Jan Albert

8:30 AM NATURAL SELECTION IN THE EVOLUTION OF SARS-COV-2 IN BATS, NOT HUMANS, CREATED A HIGHLY CAPABLE HUMAN PATHOGEN, *Sergei Pond

8:45 AM CHARACTERISING SARS-COV-2 WITHIN-HOST DIVERSITY AND IMPLICATIONS FOR RESOLVING TRANSMISSION, *Matthew Hall

9:00 AM DECOMPOSING THE SOURCES OF SARS-COV-2 FITNESS VARIATION IN THE UNITED STATES, *David Rasmussen

CIRCULATING LINEAGES OF SARS-CoV-2, *Luz Angela Alonso-Morales 9:30 AM BREAK

AGENDA

Session 3 - Modeling SARS-CoV-2 epidemic control Session Chairs: Dan Reeves

9:45 AM DRIVERS OF RESURGENT COVID-19 EPIDEMICS IN THE UNITED-STATES: SITUATION ANALYSIS THROUGH AUGUST 23 BASED ON AGE-SPECIFIC MORTALITY DATA AND AGE- SPECIFIC MOBILITY DATA AT STATE LEVEL, *Melanie Monod

10:00 AM QUANTIFYING POPULATION CONTACT PATTERNS IN THE UNITED STATES DURING THE COVID-19 PANDEMIC, *Ayesha Mahmud

10:15 AM A SIMPLE CRITERION TO DESIGN OPTIMAL NON-PHARMACEUTICAL INTERVENTIONS FOR EPIDEMIC OUTBREAKS, *Marco Tulio Angulo

10:30 AM AGE-STRUCTURED NON-PHARMACEUTICAL INTERVENTIONS FOR OPTIMAL CONTROL OF COVID-19 EPIDEMIC, *Ramses Djidjou-Demasse

10:45 AM TESTING, TRACING AND ISOLATION IN COMPARTMENTAL MODELS, *Jasmina Panovska-Griffiths

11:00 AM BREAK

Session 4 - Forecasting SARS-CoV-2 spread Session Chairs: Ruian Ke & Ramsès Djidjou-Demasse

11:15 AM REAL-TIME PROJECTIONS OF COVID-19 IN THE UNITED STATES, *Sen Pei

11:27 AM PROJECTING COVID-19 SPREAD AND MORTALITY USING RIDGE REGRESSION, *Sabrina Corsetti

11:39 AM PREDICTABILITY: CAN THE TURNING POINT AND END OF AN EXPANDING EPIDEMIC BE PRECISELY FORECAST WHILE THE EPIDEMIC IS STILL SPREADING?, *Mario Castro

11:51 AM THE RISK FOR A SECOND WAVE, AND HOW IT DEPENDS ON R0, THE CURRENT IMMUNITY LEVEL AND PREVENTIVE MEASURES, *Tom Britton

12:03 PM THE EFFECT OF EVICTIONS ON THE TRANSMISSION OF SARS-COV-2,*Justin Sheen 12:15 PM CLOSING REMARKS

Tuesday, October 20, 2020

6:00 AM Welcome and Introductions

Session 5 - Modeling SARS-CoV-2 vaccination Session Chairs: Jasmina Panovska-Griffiths

6:10 AM THE IMPACT OF COVID-19 RACIAL AND ETHNIC DISPARITIES ON HERD IMMUNITY THRESHOLDS AND OPTIMAL VACCINE ALLOCATION STRATEGIES, *Kevin Ma

6:25 AM A VACCINE THAT IS NOT FULLY PROTECTIVE, BUT INDUCES A SMALL REDUCTION IN PEAK SARS-COV-2 VIRAL LOAD AMONG INFECTED PEOPLE, COULD DECREASE INFECTIOUSNESS AND REDUCE INCIDENT INFECTIONS AND DEATHS, *Josh Schiffer AGENDA

6:40 AM DIFFERENTIAL SOCIAL DISTANCING AND SARS-CoV-2 VACCINE HESITANCY BY INFLUENZA VACCINATION STATUS IN THE UNITED STATES, *Ben Rader

6:55 AM EXPLORING POSSIBLE SCENARIOS FOR THE COVID-19 VACCINATION PROGRAM IN JAPAN, *Sung-Mok Jung

7:10 AM MODEL-INFORMED COVID-19 VACCINE PRIORITIZATION STRATEGIES BY AGE AND SEROSTATUS, *Kate Bubar

7:25 AM DYNAMIC PRIORITIZATION OF SCARCE COVID 19 VACCINES, *Jack Buckner 7:40 AM BREAK

Session 6 - Phylogenetics of SARS-CoV-2 Around the World Session Chairs: Aine O’Toole & Danielle Miller

7:55 AM THE EMERGENCE OF SARS-COV-2 IN EUROPE AND NORTH AMERICA, *Mike Worobey

8:10 AM TRACKING THE INTRODUCTION AND SPREAD OF SARS-COV-2 IN COASTAL KENYA *George Githinji

8:25 AM INSIGHTS INTO THE EVOLUTION AND EPIDEMIC SPREAD OF SARS-COV-2 IN SWITZERLAND THROUGH A LARGE-SCALE SEQUENCING EFFORT, *Tanja Stadler

8:40 AM PHYLOGENETIC ANALYSIS OF THE TIMING OF SARS-COV-2 INTRODUCTIONS INTO WASHINGTON STATE,*Diana Tordoff

8:55 AM BREAK

Session 7 - Bioinformatics & Analysis Methods for SARS-CoV-2 Session Chairs: Denise Kuehnert

9:10 AM RAPID GENOME-BASED ESTIMATION OF SARS-COV-2 INCIDENCE, *Max von Kleist

9:20 AM STABILITY OF SARS-COV-2 PHYLOGENIES, *Bryan Thornlow

9:30 AM IDENTIFICATION OF SARS-COV-2 RECOMBINANT GENOMES,*David VanInsberghe

9:40 AM OPTIMIZING VIRAL GENOMES SUBSAMPLING BY GENOMIC DIVERSITY AND TEMPORAL DISTRIBUTION, *Simone Marini

9:50 AM ULTRAFAST SAMPLE PLACEMENT ON EXISTING TREES (USHER) ENPOWERS REAL-TIME PHYLOGENETICS FOR THE SARS-COV-2 PANDEMIC, *Yatish Turakhia

10:00 AM VIRALMSA: MASSIVELY SCALABLE REFERENCE-GUIDED MULTIPLE SEQUENCE ALIGNMENT OF VIRAL GENOMES, *Niema Moshiri 10:10 AM BREAK

AGENDA

Session 8 - Modeling within-host dynamics of SARS-CoV-2 Session Chairs: Libin Rong & Mary Bushman

10:25 AM SLIGHT REDUCTION IN SARS-COV-2 EXPOSURE VIRAL LOAD DUE TO MASKING RESULTS IN SIGNIFICANT REDUCTION IN TRANSMISSION WITH WIDESPREAD IMPLMENTATION *Ashish Goyal

10:45 AM KINETICS OF SARS-COV-2 INFECTION IN THE HUMAN UPPER AND LOWER RESPIRATORY TRACTS AND THEIR RELATIONSHIP WITH INFECTIOUSNESS, *Ruian Ke

11:00 AM MODELING SARS-COV-2 VIRAL KINETICS AND ASSOCIATION WITH MORTALITY IN HOSPITALIZED PATIENTS RESULTS FROM THE FRENCH COVID-19, *Guillame Lingas

11:15 AM RKI_COVIDTESTCALCULATOR: A STANDALONE GUI TO ASSESS TESTING- AND QUARANTINE STRATEGIES FOR INCOMING TRAVELERS, CONTACT PERSON MANAGEMENT AND DE-ISOLATION, *Max Von Kleist & Wiep van der Toor

11:30 AM A PROTOTYPE QSP MODEL OF THE IMMUNE RESPONSE TO SARS-CoV-2 FOR COMMUNITY DEVELOPMENT, *Rohit Rao

11:45 AM A COALITION-DRIVEN EFFORT TO MODEL SARS-COV-2 SPREAD AND IMMUNE RESPONSE IN TISSUE, *Paul Macklin

12:00 PM SYSTEMIC AND TISSUE-LEVEL QUANTITATIVE MODELLING DISTINGUISHES THE IMMUNE RESPONSE TO SARS-COV-2, *Adrianne Jenner 12:15 PM CLOSING REMARKS THE ROLE OF BEHAVIOR, MOBILITY, AND SOCIAL NETWORKS IN SHAPING THE COVID-19 PANDEMIC *Samuel Scarpino

The COVID-19 pandemic has upended our societies and re-shaped the way we go about our day-to-day lives—from how we work and interact to the way we buy groceries and attend school. In this talk, I will present a series of studies exploring how our behavior, mobility patterns, and social networks have altered and been altered by COVID-19. Leveraging global data sets that represent billions of people, I will show how taking a complex systems approach reveals both the drivers of COVID-19 outbreaks and suggests balanced policy interventions. Then, I will discuss work by Global.health, a new collaborative network of researchers, technologists, and public health experts that has developed and built an open access platform for collecting, storing, securing, and sharing anonymized, individual-level COVID-19 data. According to Steven Johnson in The New York Times Magazine, the data captured by Global.health paints what "may well be the single most accurate portrait of the virus’s spread through the human population in existence." Only by understanding the successes and failures of our response to COVID-19 can we end this pandemic and prevent future infectious disease outbreaks from causing similar devastation.

LOCALIZED END-OF-OUTBREAK DECLARATION FOR COVID-19: EXAMPLES FROM CLUSTERS IN JAPAN

*Natalie M. Linton,1,3 Andrei R. Akhmetzhanov,2 Hiroshi Nishiura3 1Graduate School of Medicine, Hokkaido University, Japan 2College of Public Health, National Taiwan University, Taiwan 3School of Public Health, Kyoto University, Japan

End-of-outbreak declarations are an important component of outbreak response as they indicate that public health and social interventions may be relaxed or lapsed. However, traditional methods for determining the end of an outbreak based on an observation period determined solely by the incubation period of an infectious disease have previously been found to insufficiently reflect transmission dynamics. As guidance continues to be developed for coronavirus disease 2019 (COVID-19) response, we offer a method for localized end-of-outbreak determination that accounts for the reporting delay for new cases. The proposed statistical modeling approach was applied during the first wave of COVID-19 in Japan for evaluation of end-of-outbreak probabilities for case clusters. We found that end-of-outbreak determination was most closely tied with the size of the outbreak. Larger outbreaks (accounting for missing onsets and underascertainment of cases) tended to reach the proscribed probability thresholds for end-of-outbreak determination at later times compared to smaller outbreaks. In addition, end-of-outbreak determination was closely related to estimates of the basic reproduction number R0 and the overdispersion parameter k. When public health measures are effective, lower R0 (less transmission on average) and larger k (lower risk of superspreading) will be in effect, and end-of-outbreak determinations can be declared with greater confidence. In addition, this application can help distinguish between local extinction and low levels of transmission. Communicating end-of-outbreak probabilities helps inform public health decision-making around the appropriate use of resources. In this presentation, we describe our method and demonstrate its application to several clusters in Japan.

Presenting author email: [email protected] IMPROVED ESTIMATION OF TIME-VARYING REPRODUCTION NUMBERS AT LOW CASE INCIDENCE AND BETWEEN EPIDEMIC WAVES

* Kris V Parag

We construct a recursive Bayesian smoother, termed EpiFilter, for estimating the time- varying effective reproduction number, R, from the incidence of an infectious disease in real time and retrospectively. Our approach borrows from Kalman filtering theory, is quick and easy to compute, generalisable, deterministic and unlike many current methods, requires no change-point or window size assumptions. We model R as a flexible, hidden Markov state process and exactly solve forward-backward algorithms, to derive R estimates that incorporate all available incidence information. This unifies and extends two popular methods, EpiEstim, which considers past incidence, and the Wallinga-Teunis method, which looks forward in time. This combination of maximising information and minimising assumptions makes EpiFilter more statistically robust in periods of low incidence, where existing methods can struggle. As a result, we find EpiFilter to be particularly suited for assessing the risk of second waves of infection, in real time. We demonstrate the improved performance of EpiFilter on various and diverse simulated examples and on a benchmark H1N1 dataset, comparing our results with EpiEstim. We showcase the utility of EpiFilter on the COVID-19 incidence curve of New Zealand, which features a potential second wave of infection.

PRE-SYMPTOMATIC TRANSMISSION OF SARS-COV-2 IN CHINA: BEFORE AND AFTER THE LOCKDOWN

*Mary Bushman1, Colin Worby2, Hsiao-Han Chang3, Moritz Kraemer4, William P. Hanage1 1Harvard T.H. Chan School of Public Health, Boston, MA, USA 2Broad Institute, Cambridge, MA, USA 3National Tsing Hua University, Hsinchu City, Taiwan 4University of Oxford, Oxford, UK

INTRODUCTION: Symptom-based COVID-19 screening measures, such as temperature checks and symptom questionnaires, have been widely adopted as schools and workplaces reopen. The impact of such measures depends on how often SARS-CoV-2 transmits before symptoms appear. Several studies have estimated the frequency of pre-symptomatic transmission, but estimates vary widely, ranging from < 1% to 80%. Much of this variation may be attributable to differences in interpersonal contact patterns, which can alter the timing of transmission events. In particular, non-pharmaceutical interventions, such as case isolation, contact tracing, and quarantine, may prevent later transmission events, leading to an increase in the relative frequency of pre-symptomatic transmission.

METHODS: We compiled serial interval data for 873 infector-infectee pairs in China, all of whom developed symptoms in January or February 2020. In late January, China implemented sweeping measures to control the spread of SARS-CoV-2, including the lockdown of Wuhan and other cities across Hubei province. We classified case pairs as pre-lockdown (n=207) or post-lockdown (n=666) and used an MCMC approach to estimate the generation interval distribution for each time period, which we subsequently used to infer the frequency of pre-symptomatic transmission before and after the lockdown.

RESULTS: The mean generation interval was reduced by almost half following the lockdown, dropping from 7.5 days to 3.9 days. The estimated frequency of pre-symptomatic transmission was 34.4% before the lockdown; after the lockdown, the frequency doubled to 71.0%. Varying model assumptions had only modest impacts on these estimates. From these results, we calculated that non-pharmaceutical interventions reduced post-symptomatic transmission by at least 78% following the lockdown.

CONCLUSIONS: The frequency of pre-symptomatic transmission in the absence of control measures was estimated to be around 30-40%, lower than many other estimates. Non-pharmaceutical interventions drastically reduced post-symptomatic transmission in China, which increased the relative frequency of pre-symptomatic transmission.

*Correspondence: [email protected]

GENOMIC EPIDEMIOLOGY APPROACH TO ESTIMATING THE SERIAL INTERVAL DISTRIBUTION

*Kurnia Susvitasari1, Caroline Colijn1

1Department of Mathematics, Simon Fraser University, Burnaby, BC, Canada

An important key variable that characterizes the spread of an infectious disease in epidemiology is the serial interval. It represents the period between symptom onset of a primary case and symptom onset of their secondary cases. The serial interval plays an essential role in outbreak studies because it allows investigators to identify how fast the disease spreads among epidemiologically linked cases. It is typically not known who infected whom in an outbreak. Here, we will rule out the potential infectors of a case using information from their genomic clusters, where we place two cases in the same cluster if their viral sequences are less than a threshold genetic distance apart. With viral sequence data, we thereby narrow our set of potential infectors of each case, and with this information we reconstruct a collection of possible transmission trees. We also consider underreporting, in which an outbreak may have a total size higher than an official count by the authorities. We estimate the serial interval distribution from underlying hidden transmission paths among potentially linked cases. To do this, we introduce a serial interval mixture model and an algorithm to estimate the model parameters. We validate the approach with simulated data and apply it to COVID-19 sequence data. In an early estimate on the COG Consortium data we find that sublineage B.1.7 has longer serial intervals than other lineages despite having lower genetic diversity than other similarly-sized groups of sequences. REVEALING THE EXTENT OF THE COVID-19 PANDEMIC IN KENYA BASED ON SEROLOGICAL AND PCR-TEST DATA

John Ojal1,2†, *Samuel P. C. Brand1,3,4† , Vincent Were5, Emelda A Okiro6,7, Ivy K Kombe1, Caroline Mburu1, Rabia Aziza3,4, Morris Ogero1, Ambrose Agweyu1, George M Warimwe1,7, Sophie Uyoga1, Ifedayo M O Adetifa1,8, J Anthony G Scott1,8, Edward Otieno1, Lynette I Ochola-Oyier1, Charles N Agoti1,9, Kadondi Kasera10, Patrick Amoth10, Mercy Mwangangi10, Rashid Aman10, Wangari Ng’ang’a11, Benjamin Tsofa1, Philip Bejon1,7, Edwine Barasa5,7, Matt. J. Keeling3#, D. James. Nokes1,3,4#.

1Kenya Medical Research Institute (KEMRI) -Wellcome Trust Research Programme (KWTRP), Kilifi, Kenya 2London school of Hygiene and Tropical Medicine (LSHTM), UK 3The Zeeman Institute for Systems Biology and Infectious Disease Epidemiology Research (SBIDER), University of Warwick, UK. 4School of Life Sciences, University of Warwick, UK. 5 Health Economics Research Unit, KEMRI-Wellcome Trust Research Programme, Nairobi, Kenya 6Population Health Unit, Kenya Medical Research Institute -Wellcome Trust Research Programme, Nairobi, Kenya; 7 Centre for Tropical Medicine and Global Health, Nuffield Department of Medicine, University of Oxford, Oxford, United Kingdom 8Department of Infectious Diseases Epidemiology, London School of Hygiene and Tropical Medicine, London, United Kingdom 9School of Public Health, Pwani University, Kenya 10Ministry of Health, Government of Kenya, Nairobi, Kenya 11Presidential Policy & Strategy Unit, The Presidency, Government of Kenya †John Ojal and Samuel Brand contributed equally. # Matt Keeling and James Nokes contributed equally

Policy makers in Africa need robust estimates of the current and future spread of SARS-CoV-2. Data suitable for this purpose are scant. We used national surveillance PCR test, serological survey and mobility data to develop and fit a county-specific transmission model for Kenya. We estimate that the SARS-CoV-2 pandemic peaked before the end of July 2020 in the major urban counties, with 34 - 41% of residents infected, and will peak elsewhere in the country within 2-3 months. Despite this penetration, reported severe cases and deaths are low. Our analysis suggests the COVID-19 disease burden in Kenya may be far less than initially feared. A similar scenario across sub-Saharan Africa would have implications for balancing the consequences of restrictions with those of COVID-19.

*Correspondence to: Samuel Brand ([email protected])

ESTIMATING COVID-19 SECONDARY ATTACK RISK IN ELDERLY CARE FACILITIES IN FRANCE * Bastien Reyné1, Christian Selinger1, Mircea T Sofonea1, Edouard Tuaillon2, Hubert Blain3, Samuel Alizon1 1. MIVEGEC, Univ. Montpellier, IRD, CNRS, Montpellier, France 2. Laboratory, Montpellier University Hospital, France 3. Department of Geriatrics, Montpellier University Hospital, Montpellier University, Montpellier, France

The infection fatality ratio of SARS-CoV-2 infection is strongly dependent on age, and older people are more at risk to develop severe symptoms. In that regard, controlling COVID-19 epidemics in care facilities for elderly people is a major issue. We performed statistical analyses on data coming from ten elderly care facilities located in the south of France and regrouping 491 residents as well as 236 medical staff, to determine important factors impacting the transmission of the virus within those facilities. We show that earlier implementation of a generalized mask-wearing policy correlates with a lower proportion of contaminated residents. We also find that a higher number of medical staff per resident is associated with a lower proportion of contamination among residents. Finally, we estimate the secondary attack risk (SAR), a key indicator of the epidemic to implement health policies, of SARS-CoV-2 infections among residents and medical staff.

* Presenting author : Bastien Reyné ([email protected]) ESTIMATING CUMULATIVE INCIDENCE OF SARS-COV-2 WITH IMPREFECT SEROLOGICAL TESTS: EXPLOITING CUTOFF-FREE APPROACHES *Judith Bouman (1), Julien Riou (2), Sebastian Bonhoeffer (1), Roland Regoes (1)

(1) Institute of Integrative Biology, ETH Zurich, Zurich, Switzerland (2) Institute of Social and Preventive Medicine (ISPM), University of Bern, Bern, Switzerland

Large-scale serological testing in the population is essential to determine the true extent of the current coronavirus pandemic. Serological tests measure antibody responses against pathogens and define cutoff levels that dichotomize the quantitative test measures into sero-positives and negatives. With the imperfect assays that are currently available to test for past SARS-CoV-2 infection, the fraction of seropositive individuals in serosurveys is a biased estimator of the cumulative incidence and is usually corrected post-hoc to account for the sensitivity and specificity. Here we use a likelihood-based inference method --- referred to as mixture model --- for the estimation of the cumulative incidence that does not require to define cutoffs by integrating the quantitative test measures directly into the statistical inference procedure. We confirm that the mixture model outperforms the methods based on cutoffs and post-hoc corrections or Bayesian frameworks leading to less variation in point-estimates of the cumulative incidence and its temporal trend. We illustrate how the mixture model can be used to optimize the design of serosurveys with imperfect serological tests. We also provide guidance on the number of control and case sera that are required to quantify the test's ambiguity sufficiently to enable the reliable estimation of the cumulative incidence. Lastly, we show how this approach can be used to identify classes of case sera with an unknown distribution of quantitative test measures that have not been used for test validation. An application that could be used to identify the fraction of asymptomatic SARS-CoV-2 infections. Our study advocates using serological tests without cutoffs, especially if they are used to determine parameters characterizing populations rather than individuals. This approach circumvents some of the shortcomings of cutoff-based methods with post-hoc corrections or Bayesian frameworks at exactly the low cumulative incidence levels and test accuracies that we are facing in COVID-19 serosurveys.

*[email protected] bioRxiv preprint doi: https://doi.org/10.1101/2020.05.28.122366. this version posted July 30, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. It is made available under a CC-BY 4.0 International license.

Natural selection in the evolution of SARS-CoV-2 in bats, not humans, created a highly capable human pathogen

Oscar A. MacLean1,#, Spyros Lytras1,#, Steven Weaver2, Joshua B. Singer1, Maciej F. Boni3, Philippe Lemey4, Sergei L. Kosakovsky Pond2,*, David L. Robertson1,*

1MRC-University of Glasgow Centre for Virus Research, Scotland, UK. 2Temple University, Institute for Genomics and Evolutionary Medicine, Philadelphia, USA. 3Center for Infectious Disease Dynamics, Department of Biology, Pennsylvania State University, University Park, PA, USA. 4Department of Microbiology, Immunology and Transplantation, Rega Institute, KU Leuven, Leuven, Belgium.

#Joint first authors.

*To whom correspondence should be addressed: [email protected], [email protected]

Abstract RNA viruses are proficient at switching host species, and evolving adaptations to exploit the new host’s cells efficiently. Surprisingly, SARS-CoV-2 has apparently required no significant adaptation to humans since the start of the COVID-19 pandemic, with no observed selective sweeps since genome sampling began. Here we assess the types of natural selection taking place in Sarbecoviruses in horseshoe bats versus SARS-CoV-2 evolution in humans. While there is moderate evidence of diversifying positive selection in SARS-CoV-2 in humans, it is limited to the early phase of the pandemic, and purifying selection is much weaker in SARS-CoV-2 than in related bat Sarbecoviruses. In contrast, our analysis detects significant positive episodic diversifying selection acting on the bat virus lineage SARS-CoV-2 emerged from, accompanied by an adaptive depletion in CpG composition presumed to be linked to the action of antiviral mechanisms in ancestral hosts. The closest bat virus to SARS-CoV-2, RmYN02 (sharing an ancestor ~1976), is a recombinant with a structure that includes differential CpG content in Spike; clear evidence of coinfection and evolution in bats without involvement of other species. Collectively our results demonstrate the progenitor of SARS-CoV-2 was capable of near immediate human-human transmission as a consequence of its adaptive evolutionary history in bats, not humans. CHARACTERISING SARS-COV-2 WITHIN-HOST DIVERSITY AND IMPLICATIONS FOR RESOLVING TRANSMISSION

Matthew Hall*1, Katrina A. Lythgoe1, Mariateresa de Cesare1,2, Luca Ferretti1, George MacIntyre-Cockett1,2, Amy Trebes2, Monique Andersson3, Newton Otecko1, Emma L. Wise4,6, Nathan Moore4, Jessica Lynch4, Stephen Kidd4, Nicholas Cortes4, Matilde Mori7, Anita Justice3, Angie Green2, M. Azim Ansari5, Lucie Abeler-Dörner1, Catrin E. Moore1, Tim E. A. Peto3, David Eyre1,3, Robert Shaw3, Peter Simmonds5, David Buck2, John A. Todd2 on behalf of OVSG Analysis Group, David Bonsall1,2, Christophe Fraser1,2, Tanya Golubchik1,2

1Big Data Institute, Nuffield Department of Medicine, University of Oxford, Old Road Campus, Oxford OX3 7FL, UK 2Wellcome Centre for Human Genetics, Nuffield Department of Medicine, NIHR Biomedical Research Centre, University of Oxford, Old Road Campus, Oxford OX3 7BN, UK 3Oxford University Hospitals NHS Foundation Trust, John Radcliffe Hospital, Headington, Oxford, OX3 9DU, UK 4Hampshire Hospitals NHS Foundation Trust, Basingstoke and North Hampshire Hospital, Basingstoke, RG24 9NA, UK 5Peter Medawar Building for Pathogen Research, University of Oxford, OX1 3SY, UK 6School of Biosciences and Medicine, University of Surrey, Guildford, GU2 7XH, UK 7School of Medicine, University of Southampton, Southampton, SO17 1BJ, UK

For an RNA virus, SARS-CoV-2 has a low evolutionary rate due to proof-reading mechanisms that limit the influx of new mutations. However, SARS-CoV-2 phylogenies include a large number of homoplasies that are hard to reconcile with this low evolutionary rate. To understand these discrepancies, we used a targeted bait-capture approach to quantify minor allele frequencies in nearly 1400 clinical samples from two UK locations. We found that a high proportion of within-host variable sites correspond to homoplasic sites in the global phylogeny and with lineage defining sites as determined using PangoLEARN, with lower read-depth samples having higher minor allele frequencies at these sites. A high proportion of homoplasies therefore likely derive from how the within-host consensus is called at diverse sites. We discuss the likely origins of this diversity, including genuine genomic and sub-genomic diversity within sampled individuals and diversity introduced by the sequencing process, and the implications for using minor allele frequencies to help resolve patterns of transmission. DECOMPOSING THE SOURCES OF SARS-COV-2 FITNESS VARIATION IN THE UNITED STATES

David A. Rasmussen1,2,*, Lenora Kepler2, Marco Hamins-Puertolas3

1 Department of Entomology and Plant Pathology, North Carolina State University, Raleigh, North Carolina, USA 2 Bioinformatics Research Center, North Carolina State University, Raleigh, North Carolina, USA 3 Biomathematics Graduate Program, North Carolina State University, Raleigh, North Carolina, USA

Controversy has surrounded the fitness effects of mutations in the SARS-CoV-2 genome almost since the beginning of the Covid-19 pandemic. Variants such as the D614G substitution in the Spike protein rapidly increased in frequency as the pandemic spread globally, yet the impact of genetic variants on transmission potential, and therefore viral fitness at the population level, remains unclear. Here we estimate population-level fitness effects for all amino acid variants and several structural variants that circulated in the United States between February and July 2020 from viral phylogenies. We use recent extensions of birth-death models that consider how multiple traits and/or features shape pathogen fitness to learn how hundreds of genetic and non-genetic features jointly shape the fitness of SARS-CoV-2, while accounting for confounding factors such as genetic background and spatiotemporal heterogeneity in transmission rates. We also estimate how much fitness variation among pathogen lineages is attributable to genetic versus non-genetic features such as human mobility. While most genetic variants were inferred to be neutral, our analysis suggests genetic variation in viral fitness is increasing with time with several previously unreported variants contributing to fitness differences.

*Email: [email protected]

THE CONTRIBUTION OF MUTATION AND SELECTION TO PARALLEL EVOLUTION IN CIRCULATING LINEAGES OF SARS-COV-2

*Luz Angela Alonso-Morales Susan F Bailey Thomas Bataillon Rees Kassen Department of Biology, University of Ottawa, Ottawa, Canada – [email protected] Department of Biology, Clarkson University, Potsdam, NY, United States Bioinformatics Research Centre, University of Aarhus, Aarhus, Denmark Department of Biology, University of Ottawa, Ottawa, Canada

The COVID-19 pandemic has become a global health and economic emergency. Understanding the factors driving SARS-CoV-2 genome evolution is essential to inform treatment and vaccine-development strategies. Parallel evolution, the repeated evolution of the same genetic changes in evolutionarily independent populations derived from a common ancestor, is often interpreted as evidence of strong natural selection. However, parallelism can be caused by other factors as well, like the combination of high mutation rate heterogeneity and large population sizes that characterize circulating strains of SARS-CoV-2. Here, we use 45,063 publicly available SARS-CoV-2 genome sequences, collected around the world from December 2019 to August 2020, to disentangle the contribution of mutation and selection to genomic parallelism in the virus. Our approach uses the distribution of counts of synonymous (S) and non-synonymous (NS) mutations in 100 bp windows along the SARS-CoV-2 genome in circulating lineages (defined using a phylogenetic based nomenclature) to investigate the evolutionary dynamics before and after the global lockdown. Using a negative binomial model-fitting approach, we use the distribution of S across the genome as a proxy for mutation rate variation along the genome; additional variation observed in the NS substitution rate is evidence selection is driving genomic evolution. Our results indicate that the relative contributions of mutation and selection to SARS-CoV-2 evolutionary dynamics have shifted over time. Early in the pandemic (Dec - Mar) dynamics were driven by mutation, with no evidence of strong selection. Later in the pandemic (Apr - Aug), evidence that both mutation and selection drive evolutionary dynamics is emerging. We are currently extending our model framework to identify genomic covariates (dS, codon usage bias, number of protein domains, etc.) that predict the observed patterns of variation in mutation and selection across the SARS-CoV-2 genome. This will further inform how mutation and selection drive genome evolution.

DRIVERS OF RESURGENT COVID-19 EPIDEMICS IN THE UNITED-STATES: SITUATION ANALYSIS THROUGH AUGUST 23 BASED ON AGE-SPECIFIC MORTALITY DATA AND AGE-SPECIFIC MOBILITY DATA AT STATE LEVEL Mélodie Monod*, Alexandra Blenkinshop, Xiaoyue Xi, Daniel Hebert, Sivan Bershan, Valerie C Bradley, Yu Chen, Helen Coupland, Sarah Filippi, Jonathan Ish-Horowicz, Martin McManus, Thomas A Mellan, Axel Gandy, Michael Hutchinson, H Juliette T Unwin, Michaela A. C. Vollmer, Sebastian Weber, Harrison Zhu, Anne Bezanson, Simon Tietze, Neil M Ferguson, Swapnil Mishra, Seth Flaxman, Samir Bhatt, Oliver Ratmann, Nora Schmit, Lucia Cilloni, Kylie E C Ainslie, Marc Baguelin, Adhiratha Boonyasiri, Olivia Boyd, Lorenzo Cattarino, Laura V Cooper, Zulma Cucunubá, Gina Cuomo-Dannenburg, Amy Dighe, Bimandra Djaafara, Ilaria Dorigatti, Sabine L van Elsland, Richard FitzJohn, Katy Gaythorpe, Lily Geidelberg, Nicholas Grassly, William D. Green, Timothy Hallett, Arran Hamlet, Wes Hinsley, Ben Jeffrey, Edward Knock, Daniel Laydon, Gemma Nedjati-Gilani, Pierre Nouvellet, Kris V Parag, Igor Siveroni, Hayley A Thompson, Robert Verity, Erik Volz, Caroline E. Walters, Haowei Wang, Yuanrong Wang, Oliver J Watson, Peter Winskill, Charles Whittaker, Patrick GT Walker, Azra Ghani, Christl A. Donnelly, Steven M Riley, Tresnia Berah, Jeffrey W Eaton, Lucy Okell and Imperial College COVID-19 Response Team Department of Mathematics, Imperial College London, London, UK Foursquare Inc. Emodo Inc. MRC Centre for Global Infectious Disease Analysis, Jameel Institute for Disease and Emergency Analytics, Imperial College London, London, UK Novartis Pharma AG, Basel, Switzerland Department of Statistics, University of Oxford, Oxford, UK NIHR Health Protection Research Unit in Healthcare Associated Infections and Antimicrobial Resistance, Imperial College London, London, UK School of Life Sciences, University of Sussex, UK Department of Laboratory Medicine and Pathology, Brown University, Providence, RI, USA

* Correspondence: [email protected]

On March 11 2020, the Word Health Organisation declared the novel coronavirus disease (COVID-19), a global pandemic. Transmission of COVID-19 is resurging in the United States and parts of Europe since mid 2020. It is unclear how non-pharmaceutical interventions, changing contact patterns, age, and other factors drive resurgent spread, and what the impact of re-opening schools on excess deaths could be.

We collected daily, age-specific COVID-19 mortality data from 42 states and metropolitan areas across the United States since March 15, 2020. Here, we analyze the new age-specific data set in combination with aggregated, age-specific mobility trends from >10 million individuals across the United States. We developed a contact and infection model that describes time-changing contact and transmission dynamics at state-level across the United States, and which extends the model of Flaxman et al (Nature, 2020).

We analyzed data from 37 states and metropolitan areas with at least 300 COVID-19-attributed deaths, in total 5,872 observation days. We are able to reproduce highly diverse epidemic trajectories across the United States without age-shifts in contact and transmission dynamics during the epidemic. We estimate that as of August 17, 62.7% [60.1%-65.1%] of COVID-19 cases in the United States originated from adults aged 20-49, while only 1.4%[0.9%-2.1%] originated from children aged 0-9. In the context of continued, community-wide transmission primarily from adults aged 20-49, our model predicts that re-opening elementary schools could lead to substantial excess deaths over a 3-month period where COVID-19 epidemics are re- surging.

The resurgent COVID-19 epidemics in the United States are driven by adults aged 20-49, and more effective interventions that target this age group are needed to avoid substantial excess deaths in areas with community-wide transmission as schools re-open.

QUANTIFYING POPULATION CONTACT PATTERNS IN THE UNITED STATES DURING THE COVID-19 PANDEMIC

Dennis Feehan1 and Ayesha Mahmud1*

1Department of Demography, University of California, Berkeley *Presenting author: Ayesha Mahmud Department of Demography 2232 Piedmont Ave, Berkeley, CA 94720 [email protected]

Abstract SARS-CoV-2 is transmitted primarily through close, person-to-person interactions. In the absence of a vaccine, interventions focused on physical distancing have been widely used to reduce community transmission. These physical distancing policies can only control the spread of SARS-CoV-2 if they are able to reduce the amount of close interpersonal contact in a population. To quantify the impact of these policies over the first months of the COVID-19 pandemic in the United States, we conducted three waves of contact surveys between March 22 and June 23, 2020. We find that rates of interpersonal contact have been dramatically reduced at all ages in the US, with an 82% (95% CI:80% – 83%) reduction in the average number of daily contacts observed during the first wave compared to pre-pandemic levels. We find that this decline reduced the reproduction number, R0, to below one in March and early April (0.66, 95% CI:0.35 – 0.88). However, with easing of physical distancing measures, we find increases in interpersonal contact rates over the subsequent two waves, pushing R0 above 1. We also find significant differences in numbers of reported contacts by age, gender, race and ethnicity. Certain demographic groups, including people under 45, males, and Black and Hispanic respondents, have significantly higher contact rates than the rest of the population. Tracking changes in interpersonal contact patterns can provide rapid assessments of the impact of physical distancing policies over the course of the pandemic and help identify at-risk populations.

A SIMPLE CRITERION TO DESIGN OPTIMAL NON-PHARMACEUTICAL INTERVENTIONS FOR EPIDEMIC OUTBREAKS Marco Tulio Angulo*a, Fernando Castañosb, Jorge X. Velasco-Hernandezc, Rodrigo Moreno- Mortond, and Jaime A. Morenoe. aCONACyT - Institute of Mathematics, National Autonomous University of Mexico, Juriquilla, 76230, Mexico bDepartment of Automatic Control, CINVESTAV-IPN, Ciudad de México, 07360, Mexico cInstitute of Mathematics, National Autonomous University of Mexico, Juriquilla, 76230, Mexico dFaculty of Sciences, National Autonomous University of Mexico, Ciudad de México, 04510, México eInstitute of Engineering, National Autonomous University of Mexico, Ciudad de México, 04510, México

Abstract. For mitigating the COVID-19 pandemic, much emphasis exists on the objective of implementing non-pharmaceutical interventions to keep the reproduction number below one. However, using that objective ignores that some of these interventions, like bans of public events or lockdowns, must be transitory and as short as possible because of their significative economic and societal costs. Here we derive a simple and mathematically rigorous criterion for designing optimal transitory non-pharmaceutical interventions for mitigating epidemic outbreaks. We find that reducing the reproduction number below one is sufficient but not necessary. Instead, our criterion prescribes the required reduction in the reproduction number according to the desired maximum of disease prevalence and the maximum reduction in disease transmission that the interventions can achieve. We study the implications of our theoretical results for designing non- pharmaceutical interventions in 16 cities and regions during the COVID-19 pandemic. In particular, we estimate the minimal reduction of each region's contact rate that is necessary to control the epidemic optimally. Our results contribute to establishing a rigorous methodology to guide the design of optimal non-pharmaceutical intervention policies.

Preprint: https://www.medrxiv.org/content/10.1101/2020.05.19.20107268v1

Correspondence to: Marco Tulio Angulo, Ph. D Institute of Mathematics National Autonomous University of Mexico Juriquilla, Mexico email: [email protected] METASTABILITY PROPERTY OF A MODEL FOR THE EVOLUTION OF A FUNGAL PATHOGEN

Quentin Richard1, Samuel Alizon1, Marc Choisy1,2,3, Mircea T. Sofonea1, Ramsès Djidjou-Demasse1,*

1MIVEGEC, Univ. Montpellier, IRD, CNRS, Montpellier, France. 2Centre for Tropical Medicine and Global Health, Nuffield Department of Medicine, University of Oxford, UK. 3Oxford University Clinical Research Unit, Ho Chi Minh City, Vietnam.

In the absence of efficient treatment or vaccine, non-pharmaceutical interventions (NPIs) have emerged as a key control issue of COVID-19 pandemics. A particularly controversial issue has to do with age-specific control measures. We tackle this issue by applying optimal control theory to an original model structured by both the age of host population and time since individuals are infected. This double continuous structure represents a major achievement from a technical standpoint, but it is also particularly relevant from a biological standpoint. Indeed, individuals widely differ in the way they spread the infection depending on their age and on the time since they have been infected. In the model, the control consists in decreasing the force of infection, as can be achieved via NPIs.

We find that, if the time until treatment of vaccine deployment is assumed to be a year, the optimal control strategy consists in a relatively strong intervention on the older populations during a hundred days, followed by a steady decrease in a way that depends on the cost associated to a such control. The intervention on the younger population is weaker and occurs only if the cost associated with the control is relatively low. A uniform constant control over the whole populations or over its younger fraction are both strongly outperformed by the optimal strategy in terms of the number of casualties. These results bring new facts to the debate about the relevance of age-based interventions to control the COVID-19 pandemics and open promising avenues of research, for instance of age-based contact tracing. [email protected] TESTING, TRACING AND ISOLATION IN COMPARTMENTAL MODELS

Simone Sturniolo1 , William Waites2,Tim Colbourn3 , David Manheim4, Jasmina Panovska- Griffiths3,5,6*

1 Science and Technology Facilities Council, Swindon, UK 2 School of Informatics, University of Edinburgh, Edinburgh, Scotland, UK 3 UCL Institute for Global Health, London, UK 4 University of Haifa Health and Risk Communication Research Center, Haifa, Israel 5 Department of Applied Health Research, UCL, London, UK 6 The Queen’s College, Oxford University, Oxford, UK

Existing compartmental mathematical modelling methods for epidemics, such as the Susceptible- Exposed-Infectious-Removed (SEIR) models, cannot accurately represent effects of contact tracing. This makes them inappropriate for evaluating testing and contact tracing strategies to contain an outbreak. An alternative used in practice is the application of agent-based models (ABMs), but these can often be complex, less well-understood and much more computationally expensive. This work showcases a new method for accurately including the effects of Testing, contact-Tracing and Isolation (TTI) strategies in standard SEIR models.

We derive our method using a careful probabilistic argument to show how contact tracing at the individual level is reflected in aggregate on the population level. Our adaptation to incorporate testing is applicable across compartmental models (e.g. SIR, SIS etc) and across infectious diseases.

We validate the novel SEIR-TTI model against a mechanistic agent-based model where testing, tracing and isolation of individuals is explicitly represented. We show that we can achieve good agreement at far less computational cost. We also showcase how the SEIR-TTI model can be applied to the COVID-19 pandemic to understand the impact of possible TTI strategy to control this outbreak. We illustrate this using a user-friendly interface we have generated and that can be found at https://app.covidtti.com/.

*corresponding author and presenter [email protected] preprint can be found at https://www.medrxiv.org/content/10.1101/2020.05.14.20101808v2.full.pdf REAL-TIME PROJECTIONS OF COVID-19 IN THE UNITED STATES

Sen Pei*, Teresa Yamana, Jeffrey Shaman Department of Environmental Health Sciences, Mailman School of Public Health, Columbia University

COVID-19 will continue circulating in the United States until effective vaccines and therapeutic treatments become widely available. Reliable real-time projections of local COVID-19 transmission can support timely decision-making on non-pharmaceutical intervention measures. Since March 2020, our group has been generating scenario-based projections of COVID-19 spread in 3142 US counties. We calibrate a dynamical metapopulation model informed by human movement across US counties to daily reported case and death numbers, and generate county-level projections for daily cases, deaths, hospitalizations and ICU admissions under a range of control scenarios. The forecasting system also provides estimation of local effective reproductive numbers. We continually update the forecasting system as new data sources become available, and with increased understanding on the epidemiological features of COVID- 19 transmission. Here I will give an overview of the model structure and calibration method, present the utility of projection results, and further discuss the challenges in generating real- time projections for an emerging infectious disease under a highly nonstationary situation.

Email: [email protected] PROJECTING COVID-19 SPREAD AND MORTALITY USING RIDGE REGRESSION *Sabrina Corsetti1, Ella McCauley1, Robert Myers1, Thomas Baer2, Tom Schwarz1 1University of Michigan, Ann Arbor, MI 2Trinity University, San Antonio, TX

Presenting Author: Sabrina Corsetti, University of Michigan, Ann Arbor, MI, 1(815)641-6924, [email protected]

This study implemented machine learning techniques for the data-based prediction of COVID- 19 case and death counts. The resulting model has been used primarily for predictions within the United States at the national and state levels, but it is generalizable to any global region. It is a weekly contributor to the COVID-19 Forecasting Hub (UMich model), which feeds into the CDC’s weekly COVID-19 predictions (UM model). The model’s absence of explicit assumptions makes it relatively unique among the available COVID-19 models. It achieves this freedom by using ridge regression – a type of machine learning algorithm – to predict future data trends solely based on previous data. The algorithm treats each day’s cases and deaths as a linear combination of a specific number of previous days’ data, with the same linear combination coefficient vector applied to each day. This coefficient vector is optimized during the training of the ridge regression algorithm and is applied repeatedly to generate future data points for up to four weeks. Accompanying the presentation of the model are several studies regarding its performance, namely: a relative comparison of performance across variations of the model (e.g., those with predictions made on a weekly versus a daily basis); an application of its predictions to case fatality rates and other pandemic characteristics; and the uncertainty and sensitivity of the linear combination coefficient vector components. Predictability: Can The Turning Point And End Of An Expanding Epidemic Be Precisely Forecast While The Epidemic Is Still Spreading?

Mario Castroa,b, Saul Aresa,c, Jose A. Cuesta a,d,e,f, and Susanna Manrubiaa,c aGrupo Interdisciplinar de Sistemas Complejos (GISC), Madrid, Spain b Instituto de Investigacion Tecnologica (IIT), Universidad Pontificia Comillas, Madrid, Spain cDept. Biologıa de Sistemas, Centro Nacional de Biotecnolog´ıa (CSIC). c/ Darwin 3, 28049 Madrid, Spain dDept. Matematicas, Universidad Carlos III de Madrid, Avenida de la Universidad 30, 28911 Leganes, Spain ´ e Instituto de Biocomputacion y Fısica de Sistemas Complejos (BIFI), c/ Mariano Esquillor, Campus Rıo Ebro, Universidad de Zaragoza, 50018 Zaragoza, Spain fUC3M-Santander Big Data Institute (IBiDat), c/ Madrid 135, 28903 Getafe, Spain

Epidemic spread is characterized by exponentially growing dynamics, which are intrinsically unpredictable. The time at which the growth in the number of infected individuals halts and starts decreasing cannot be calculated with certainty before the turning point is actually attained; neither can the end of the epidemic after the turning point. An SIR model with confinement (SCIR) illustrates how lockdown measures inhibit infection spread only above a threshold that we calculate. The existence of that threshold has major effects in predictability: A Bayesian fit to the COVID-19 pandemic in Spain shows that a slow-down in the number of newly infected individuals during the expansion phase allows to infer neither the precise position of the maximum nor whether the measures taken will bring the propagation to the inhibition regime. There is a short horizon for reliable prediction, followed by a dispersion of the possible trajectories that grows extremely fast. The impossibility to predict in the mid-term is not due to wrong or incomplete data, since it persists in error-free, synthetically produced data sets, and does not necessarily improve by using larger data sets. Our study warns against precise forecasts of the evolution of epidemics based on mean-field, effective or phenomenological models, and supports that only probabilities of different outcomes can be confidently given. THE RISK FOR A SECOND WAVE, AND HOW IT DEPENDS ON R0, THE CURRENT IMMUNITY LEVEL AND PREVENTIVE MEASURES,

*Tom Britton

The COVID-19 pandemic has hit different parts of the world differently. Some regions are still in the rise of the first wave, other regions have experienced a first wave but are now facing a decline, and other regions have already starting to see a second wave. The cumulative fraction of infected individuals in the different regions primarily depend on two factors: a) the initial potential for COVID-19 in the region (often quantified with the basic reproduction number $R_0$), and b) the timing, amount and effectiveness of preventive measures put in place. Regions differ in both these respects thus leading to different immunity levels in the regions. The highly important question addressed here is an aim to quantify the amount of future restrictions necessary to avoid a new big epidemic wave, given the initial $R_0$ and the current immunity level $\hat i$. This is done for an epidemic model allowing for individual heterogeneities with respect to susceptibility, age-cohorts and social activity levels, and which is calibrated to a specific region by fitting $R_0$ and the current immunity level to it. THE EFFECT OF EVICTIONS ON THE TRANSMISSION OF SARS-CoV-2. *Justin K. Sheen ([email protected]), ; Michael Z. Levy, University of Pennsylvania; Alison L. Hill, Harvard University; Anjalika Nande, Harvard University; Andrew J. Greenlee, University of Illinois; Daniel W. Schneider, University of Illinois; Emma L. Walters, University of Illinois; M. Florencia Tejeda, University of Pennsylvania.

The COVID-19 pandemic has led to surges in unemployment, which in turn have left many tenants unable to pay their rent. This looming eviction crisis could have severe consequences on SARS-CoV-2 transmission. Studies show that many evicted households “double-up,” moving ​ in with family or friends. Doubling-up shifts the distribution of household sizes upwards, facilitating SARS-Cov-2 spread. To study the effect of evictions on SARS-CoV-2 at a population level, we present an SEIR network model tracking both household and external contacts, and ​ use it to simulate epidemic trajectories within a theoretical metropolitan area of one million individuals. We divide the simulation into temporal phases to reproduce the epidemic since early 2020: (1) an early exponential phase, followed by (2) lockdown of external contacts, (3) relaxation of lockdown of external contacts. In the fourth phase we project the course of the epidemic beyond Fall 2020 under two counterfactual scenarios, one in which evictions continue, and the other in which they are halted. In some simulations we consider a fifth phase in which lockdown is reinitiated. In a simple model, which assumes homogeneous mixing in the simulated city, we expect a 1%/month eviction rate (with all evictions resulting in doubling-up) to cause a 3-6% increase in seroprevalence vs. if evictions were prevented, and approximately 1 excess death for every 70 evicted households. The effect of evictions is even more profound in models that consider clustering of evictions and transmission in poorer neighborhoods. We also consider evictions that lead to homelessness, and relocation to shelters/encampments. Across all scenarios, we find that evictions (1) increase the total number of infected individuals, in both evicted and non-evicted households; and, (2) decrease the efficacy of a hypothetical second lockdown. The generality of our results provide a theoretical basis to assess eviction moratoriums in any city. THE IMPACT OF COVID-19 RACIAL AND ETHNIC DISPARITIES ON HERD IMMUNITY THRESHOLDS AND OPTIMAL VACCINE ALLOCATION STRATEGIES Kevin C. Ma1,*, Tigist F. Menkir2, Stephen Kissler1, Yonatan H. Grad1,3, Marc Lipsitch2

1Department of Immunology and Infectious Diseases, Harvard T.H. Chan School of Public Health, Boston, USA

3Division of Infectious Diseases, Brigham and Women’s Hospital and Harvard Medical School, Boston, USA

2Center for Communicable Disease Dynamics, Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, USA

Understanding the factors that influence herd immunity to SARS-CoV-2 can inform policies for controlling the pandemic. Models incorporating heterogeneity in exposure and connectivity result in lower herd immunity thresholds (HITs) compared to homogeneous models, but the social factors that govern heterogeneity remain unknown. Here, we fit compartmental models structured by race and ethnicity to seroprevalence data from New York State to assess the impact on HITs and final epidemic sizes. We also use census estimates of the interactions between racial and ethnic groups to inform the degree of assortativity. Allowing exposure levels to vary by race and ethnicity in a proportionate mixing model resulted in HITs of 28% in Long Island and 42% in New York City compared to 50% for a homogenous model with R0 = 2. However, increased assortativity — i.e., preferential mixing within groups — returned the herd immunity threshold to values closer to the well-mixed model: with 80% of contacts exclusively within-group, HITs increased to 44% in Long Island and 49% in New York City. Using social mixing matrices informed by census data, we assessed the impact of different vaccine distribution frameworks, finding that preferential vaccination of high-risk demographic groups resulted in vaccine- conferred herd immunity at substantially lower numbers of vaccines administered compared to random vaccination. Overall, our findings highlight the importance of modeling racial and ethnic disparities when estimating disease-induced and vaccine-conferred herd immunity levels for SARS-CoV-2.

*Presenting author. Correspondence to [email protected] A VACCINE THAT IS NOT FULLY PROTECTIVE, BUT INDUCES A SMALL REDUCTION IN PEAK SARS-COV-2 VIRAL LOAD AMONG INFECTED PEOPLE, COULD DECREASE INFECTIOUSNESS AND REDUCE INCIDENT INFECTIONS AND DEATHS

Ashish Goyal1†, Dave Swan1†, Elizabeth Krantz1, Mia Moore1, Peter Gilbert1,2, Holly Janes1,2, Fei Gao1, Fabian Cardozo-Ojeda1, Daniel B. Reeves1, Lawrence Corey1,3,4,5, Bryan T. Mayer1, Dobromir Dimitrov1††, Joshua T Schiffer1,4,5††

1 Vaccine and Infectious Diseases Division, Fred Hutchinson Cancer Research Center 2 Department of Biostatistics, University of Washington, Seattle 3 Department of Laboratory Medicine, University of Washington, Seattle 4 Department of Medicine, University of Washington, Seattle 5 Clinical Research Division, Fred Hutchinson Cancer Research Center

† These authors contributed equally to the work. †† These authors contributed equally to the work.

The primary efficacy endpoint in all ongoing SARS-CoV-2 phase 3 vaccine trials is reduction in symptomatic disease (VEDIS) which reflects a combination of reduction in susceptibility to infection (VES) and reduction in percent symptomatic following acquisition (VEP). VEDIS=50% (e.g. with VES=40% and VEP=17%) is the benchmark set for a minimally effective vaccine. Further benefit may be accrued at the population level due to reduction in infectiousness upon acquisition (VEI). We developed a SARS-CoV-2 transmission model and calibrated it to COVID-19 cases, deaths and hospitalizations in King County, Washington. We modeled implementation of vaccines to 5000-10000 people per day over 6 months, assuming current transmission conditions. A vaccine with VEDIS=50% (VES=40%, and VEP=17%) and VEI=0% would potentially lower cases and deaths by 30-50% over the subsequent year, while also allowing significant relaxation of physical distancing. A vaccine with VEI=50% and VEDIS=0.19 (VES=10% and VEP=10%) would achieve comparable reductions assuming equivalent rates of implementation.

Assessment of VEI is challenging for SARS-CoV-2 because most transmissions occur during the pre- symptomatic phase of infection when viral load is highest. To identify the link between viral load reduction and the probability of transmission at the individual and population level, we used a separate viral dynamics model that simulates contacts between potential transmitters and exposed contacts. We identified that a vaccine induced 0.9 log reduction in peak viral load would allow VEI=50%, and that a 2.5 log reduction in peak viral load would allow VEI=90%.

Ongoing phase 3 vaccine efficacy trials will only capture viral load following symptoms in COVID-19 cases, and thus will not provide information on peak viral load which occurs in the pre-symptomatic period; VEI is therefore not evaluable. To inform future vaccine trials in which evaluating VEI is a primary or key secondary objective, we simulated different frequencies of nasal sampling to diagnose incident infection and measure post-infection viral load. We identified that VEI can be most accurately estimated if sampling occurs daily. A vaccine trial with a total of 76 breakthrough infections divided equally between the vaccine and placebo arms (due to low VES) could be adequately powered to detect a vaccine effect on viral load that is predicted to translate to VEI>50%. Because moderate VEI could lead to large population reductions in incident cases and deaths, immunogenic vaccines - including those that do not achieve predetermined goals for VEDIS in ongoing phase 3 trials - should be considered for clinical trials which sample frequently to diagnose infection and capture post-infection viral load. DIFFERENTIAL SOCIAL DISTANCING AND SARS-CoV-2 VACCINE HESITANCY BY INFLUENZA VACCINATION STATUS IN THE UNITED STATES *Rader B1,2, Burns MR1,3, Hawkins JB1,4, Sewalk K1, Brownstein JS1,4 1Boston Children’s Hospital, 2Boston University School of Public Health, 3University of Virginia, 4Harvard Medical School

Introduction: Human-to-human transmission of SARS-CoV-2 is facilitated by close contact between persons. Social distancing is meant to limit the spread by reducing contacts; however, there exists substantial heterogeneity in the uptake of this nonpharmaceutical intervention across geospatial, political and socioeconomic groups. Here we investigate differential contacts by influenza vaccine status. Further, we connect influenza vaccination to SARS-CoV-2 vaccine hesitancy.

Methods: Responses [n=504,387] across two surveys between April and July 2020 were collected via a joint SurveyMonkey.com and COVIDNearYou.org web platform. Individuals self-reported on whether they received an influenza vaccination in the last year, their average number of close (within 6ft) daily contacts, personal demographics, and their home state. A second survey queried individuals’ hesitancy to receive a future SARS-CoV-2 vaccine. A mixed model with a random intercept for state was fit to assess the association between influenza vaccination status with close daily contacts and influenza vaccination status with SARS-CoV-2 vaccine hesitancy. The association between self-reported contacts and open-source Google mobility measures was also assessed.

Results: Adjusting for age, sex, race, and income, those who were previously vaccinated against influenza reported 4.0 [3.1 - 5.0] fewer daily contacts than those not vaccinated. This disparity remained stable across the study period. Substantial variation [r2 = .45] in mobility reports was explained by aggregated survey results, suggesting self-reports approximate community-level mobility. Additionally, those not vaccinated for influenza reported 0.79 [0.76 - 0.82] points higher on a 5-point SARS-CoV-2 vaccine-hesitancy scale, indicating increased skepticism.

Discussion: Efficient vaccine targeting is essential to disrupt the transmission of SARS-CoV-2. We show that individuals not vaccinated for influenza have contact patterns that put them at higher risk for SARS- CoV-2 infection and propagation, as well as report higher vaccine hesitancy. Mathematical models and future vaccination campaigns need to account for these disparities and potential increased difficulty reaching these individuals.

*Presenting Author ([email protected]) EXPLORING POSSIBLE SCENARIOS FOR THE COVID-19 VACCINATION PROGRAM IN JAPAN

*Sung-mok Jung1,2, Andrei R. Akhmetzhanov2, Natalie M. Linton1,2 and Hiroshi Nishiura1,3 1Kyoto University School of Public Health, Kyoto, Japan 2Graduate School of Medicine, Hokkaido University, Sapporo, Hokkaido, Japan 3CREST, Japan Science and Technology Agency, Saitama, Japan

As the Japanese government established an agreement with the pharmaceutical companies to supply the country with a coronavirus disease (COVID-19) vaccine, to plan for an efficient vaccination program in Japan, the required amount of vaccine and the first priority age-group has been debated. Here, we quantified age-dependent next generation matrix and estimated the impact of possible scenarios for the COVID-19 vaccination program on the effective reproduction number (R) and the cumulative number of deaths, employing a mathematical model and analyzing growth phage incidence of Tokyo metropolis and Hokkaido prefecture. The R of the first wave of disease was estimated as 1.63 (1.43–1.84) in Tokyo and as 1.31 (1.15–1.50) in Hokkaido, respectively. The model using the constructed next generation matrix indicated that vaccinating those aged from 20–59 using at least 40 million doses and those aged from 50–80 years and older with more than 30 million doses is required to keep R < 1 in Tokyo and Hokkaido, while vaccinating adults aged from 60–79 years can lead to the substantial reduction in the cumulative number of deaths in both locations. The proposed model identified the priority age- group for the COVID-19 vaccination program, with respect to the achievement of herd immunity and minimizing fatal cases in Japan.

Correspondence: Sung-mok Jung ([email protected]) MODEL-INFORMED COVID-19 VACCINE PRIORITIZATION STRATEGIES BY AGE AND SEROSTATUS Kate M. Bubar*, Stephen M. Kissler, Marc Lipsitch, Sarah Cobey, Yonatan H. Grad, Daniel B. Larremore

*Corresponding email: [email protected]

Department of Applied Mathematics, University of Colorado Boulder IQ Biology Program, University of Colorado Boulder Department of Immunology and Infectious Diseases, Harvard T.H. Chan School of Public Health Center for Communicable Disease Dynamics, Harvard T.H. Chan School of Public Health Department of Ecology and Evolution, University of Chicago Department of Computer Science, University of Colorado Boulder BioFrontiers Institute, University of Colorado Boulder

When a vaccine for COVID-19 becomes available, limited initial supply will raise the question of how to prioritize the available doses and thus underscores the need for transparent, evidence-based strategies that relate knowledge of, and uncertainty in, disease transmission, risk, vaccine efficacy, and existing population immunity. Here, we employ a model-informed approach to vaccine prioritization that evaluates the impact of prioritization strategies on cumulative incidence and mortality and accounts for population factors such as age, contact structure, and seroprevalence, and vaccine factors including imperfect and age-varying efficacy. This framework can be used to evaluate and compare existing strategies, and it can also be used to derive an optimal prioritization strategy to minimize mortality or incidence. We find that a transmission-blocking vaccine should be prioritized to adults ages 20-49y to minimize cumulative incidence and to adults over 60y to minimize mortality. Direct vaccination of adults over 60y minimizes mortality for vaccines that do not block transmission.We also estimate the potential benefit of using individual-level serological tests to redirect doses to only seronegative individuals, improving the marginal impact of each dose. We argue that this serology-informed vaccination approach may improve the efficiency of vaccination efforts while partially addressing existing inequities in COVID-19 burden and impact.

DYNAMIC PRIORITIZATION OF SCARCE COVID 19 VACCINES

*Jack Buckner, Gerardo Chowell and Michael Springborn

Graduate group in ecology, University of california, Davis Department of Population Health Sciences, School of Public Health, Georgia State University Environmental Science and Policy, University of California, Davis

*Email: [email protected] *Phone: 206-484-2444

Abstract: Multiple promising COVID-19 vaccines are under rapid development, with deployment of the initial supply expected in late 2020 or early 2021. Achieving an objective vaccine prioritization scheme across socio-demographic groups is an imminent and crucial public policy challenge given that (1) the eventual vaccine supply will be highly constrained for at least several months after launching a vaccination campaign and (2) there are stark differences in transmission and severity impact of SARS- CoV-2 across groups. We assess the optimal allocation of a limited and dynamic COVID-19 vaccine supply in the U.S. across socio-demographic groups differentiated by age and essential worker status. The transmission dynamics are modeled using a compartmental (SEIR) model parameterized to capture our current understanding of the transmission and epidemiological characteristics of COVID-19, including key sources of group heterogeneity (susceptibility, severity, and contact rates). We investigate tradeoffs between three alternative policy objectives: minimizing infections, years of life lost, or deaths. Moreover, we model dynamic vaccine prioritization policies that respond to changes in the epidemiological status of the population as SARS-CoV-2 continues its march. Because contacts tend to be concentrated within age groups, there is diminishing marginal returns as vaccination coverage increases in a given group, increasing its protective immunity against infection or severe disease outcomes. We find that optimal prioritization consistently targets older essential workers. However, depending on the policy objective, younger essential workers are prioritized to minimize infections or those over 59 years old in order to minimize mortality burden. We also find that optimal strategies significantly outperform uniform vaccination strategies by up to 24% depending on the outcome optimized. For example, in our baseline model, cumulative mortality decreased on average by 16% (25,000 deaths in the U.S. population) over the course of the outbreak. THE EMERGENCE OF SARS-COV-2 IN EUROPE AND NORTH AMERICA

*Michael Worobey1, Jonathan Pekar2,3, Brendan B. Larsen1, Martha I. Nelson4, Verity Hill5, Jeffrey B. Joy,6,7,8, Andrew Rambaut5, Marc A. Suchard9,10,11, Joel O. Wertheim12, Philippe Lemey13

1Department of Ecology and Evolutionary Biology, University of Arizona, Tucson, AZ 85721, USA. 2Bioinformatics and Systems Biology Graduate Program, University of California San Diego, La Jolla, CA 92093, USA 3Department of Biomedical Informatics, University of California San Diego, La Jolla, CA 92093, USA 4Fogarty International Center, National Institutes of Health, Bethesda, Maryland 20892, USA. 5Institute of Evolutionary Biology, University of Edinburgh, King's Buildings, Edinburgh, EH9 3FL, UK. 6Department of Medicine, University of British Columbia, Vancouver, BC, Canada 7BC Centre for Excellence in HIV/AIDS, Vancouver, BC, Canada 8Bioinformatics Programme, University of British Columbia, Vancouver, BC 9Department of Biomathematics, David Geffen School of Medicine, University of California Los Angeles, Los Angeles, CA 90095, USA 10Department of Biostatistics, Fielding School of Public Health, University of California Los Angeles, Los Angeles, CA 90095, USA 11Department of Human Genetics, David Geffen School of Medicine, University of California Los Angeles, Los Angeles, CA 90095, USA 12Department of Medicine, University of California San Diego, La Jolla, CA 92093, USA 13KU Leuven Department of Microbiology, Immunology and Transplantation, Rega Institute, Laboratory of Clinical and Evolutionary Virology, Leuven, Belgium

Accurate understanding of the global spread of emerging viruses is critically important for public health response and for anticipating and preventing future outbreaks. Here, we elucidate when, where and how the earliest sustained SARS-CoV-2 transmission networks became established in Europe and North America. Our results suggest that rapid early interventions successfully prevented early introductions of the virus into Germany and the US from taking hold. Other, later introductions of the virus from China to both Italy and to Washington State founded the earliest sustained European and North America transmission networks. Our analyses demonstrate the effectiveness of public health measures in preventing onward transmission and show that intensive testing and contact tracing could have prevented SARS-CoV-2 from becoming established.

*Presenter Email: [email protected]

Tracking the introduction and spread of SARS-CoV-2 in Coastal Kenya *George Githinji1, D. James Nokes1,2, Charles N. Agoti1,3

Affiliations 1 KEMRI-Wellcome Trust Research Programme, Kilifi, Kenya 2School of Life Sciences and Zeeman Institute for Systems Biology and Infectious Disease Epidemiology Research (SBIDER), University of Warwick, Coventry, United Kingdom. 3School of Public Health, Pwani University, Kenya Correspondence [email protected]

There are limited SARS-CoV-2 genomic data from East-Africa. We generated 274 SARS-CoV-2 genomes collected from coastal Kenya between March and June 2020. We identified eight global lineages and at least 76 independent SARS-CoV-2 introductions into the region. The B.1 lineage accounted for 82.1% of the 274 sequenced cases. Lineages A, B and B.4 were detected only in samples collected during screening at the Kenya-Tanzania border or of travellers from Dubai. Lineage B.1.1.33 was detected in a single sample collected during mass testing in Mombasa. Despite the introduction of multiple lineages in the region, none showed extensive local expansion. Our data suggest that SARS-CoV-2 introductions to coastal Kenya were predominantly of European origin. We speculate that the Kenya-Tanzania border and the major seaport and international air terminal in Mombasa were important sources of SARS-CoV-2 importations. The established transmission in coastal Kenya is largely due to viruses from the B.1 lineage, which appears to have been introduced on at least 45 occasions. Additional sequences will provide insights on novel mutations from local and community transmission patterns. PHYLOGENETIC ANALYSIS OF THE TIMING OF SARS-COV-2 INTRODUCTIONS INTO WASHINGTON STATE

*Diana M. Tordoff1,3, Joshua T. Herbeck2,3

1 University of Washington, Department of Epidemiology, Seattle, WA 2 University of Washington, Department of Global Health, Seattle, WA 3 International Clinical Research Center, University of Washington, Department of Global Health, Seattle, WA

Background: The first confirmed case of SARS-CoV-2 in Washington state (WA) was identified on January 21, 2020. We aimed to estimate the number and timing of introductions of SARS-CoV-2 lineages in WA.

Methods: Our analysis used full genome SARS-CoV-2 sequences from GISAID sampled between December 2019 and June 2020. Sequences were aligned using MAFFT and phylogenetic trees were estimated using IQTREE, by pangolin lineage (A, B, B.1, and B.1.1). We generated 5 replicates that each included 1802 high quality WA sequences and 3500 non-WA sequences, including 2970 non-WA sequences that were closest to WA sequences based on the raw number of mutations and a time- stratified random sample of 530 additional non-WA sequences. In order to incorporate phylogenetic uncertainty into our estimates we time calibrated each phylogeny 100 times for 100 random polytomy resolutions using the treedater algorithm and assuming a strict molecular clock. Maximum parsimony ancestral state reconstruction was then used to estimate the state (WA or non-WA) of each node. Internal node date for the MRCA of each WA subclade was assumed to be the date of SARS-CoV-2 introduction.

Results: We estimated a mean of 28 separate introductions (range 12-50) of SARS-CoV-2 into WA through June 2020, beginning in mid-January 2020 and peaking in number on March 29, 2020. The resulting WA subclades ranged in size from 2-912 sequences: 36% were cherries while 24% of WA subclades included 20 or more sequences, suggestive of local transmission chains. The majority of introductions were lineage B.1 (56%), followed by lineage A (16%), B (14%) and B.1.1 (14%).

Conclusions: We found phylogenetic evidence that the SARS-CoV-2 epidemic in WA was seeded by multiple ongoing introductions, although due to incomplete sampling our estimates underestimate the number of introductions. Our findings suggest that imported SARS-CoV-2 lineages decreased following WA’s stay-at-home order on March 23, 2020 with a one-week lag.

Corresponding author: Diana M. Tordoff, [email protected] INSIGHTS INTO THE EVOLUTION AND EPIDEMIC SPREAD OF SARS-COV-2 IN SWITZERLAND THROUGH A LARGE-SCALE SEQUENCING EFFORT

*Tanja Stadler

Europe was the epi-center of the SARS-CoV-2 pandemic in spring of 2020 and is currently experiencing a second increase in confirmed cases. Switzerland is in the center of Europe and well-connected to the rest of the European countries. Thus we hypothesize that viral import dynamics and epidemiological dynamics within the country are exemplary for a European country. Here, we quantify evolutionary and epidemiological dynamics of SARS-CoV-2 in Switzerland based on genomic data. In a nation-wide sequencing effort, we produced >1400 SARS-CoV-2 genomes for the first 6 months of the Swiss epidemic. We show that the SARS-CoV-2 population currently circulating in Switzerland is not much different from the virus population in March. We then show that the Swiss epidemic was initially driven by imported strains from our neighboring countries while current imports mainly stem from Belgium and the Netherlands. The relative importance of import versus local transmission became smaller through time, which in turn shows that contact tracing is becoming increasingly important to break transmission chains. Finally, we quantify the effective reproductive number solely based on genomic data confirming estimates based on line list data. Overall, we show how our understanding of an epidemic is improved by considering viral genomic data.

RAPID GENOME-BASED ESTIMATION OF SARS-COV2 INCIDENCE

Maureen Rebecca Smith1,2+, Maria Trofimova1,2+, Yannick Duport1,2, Ariane Weber3, Denise Kühnert3,4, *Max von Kleist1,2,4, on behalf of the working group on SARS-CoV-2 Molecular Surveillance at RKI and the GisAID EpiCoV group

1,2 Systems Medicine of Infectious Disease (P5) and Bioinformatics MF1 Department, Robert Koch Institute Berlin, Germany 3 Transmission, Infection, Diversification and Evolution Group, Max-Planck Institute for the Science of Human History, Jena, Germany 4 German COVID Omics Initiative (deCOI)

+contributed equally *presenter; [email protected]

By mid-September 2020, over 28 million individuals have tested positive for the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and more than 900.000 deaths were associated with the infection. However, the true number of infections is unknown and believed to exceed the reported numbers by 1.3-100 fold. PCR-based testing is currently the gold standard for monitoring and to guide the definition of risk areas for which mobility and travel restrictions are required. However, temporal changes in testing availability and differences in testing policies strongly affect the proportion of undetected cases. To overcome this testing bias and better assess SARS-CoV-2 transmission dynamics, we propose a genome-based method to estimate the number of missing cases.

Our method assigns SARS-CoV2 sequences to temporal bins and uses the “number of independent origins” [1] to approximate spreading dynamics. Different binning strategies result in point clouds, through which we fit a weighted smoothing spline that incorporates statistical uncertainty.

The method performs robustly in a range of simulated population dynamics- and evolution scenarios. It reliably reconstructs epidemiological dynamics, even with low sampling proportions and for multiple independent introductions. We applied the method to GisAID and in-house sequencing data to estimate incidence in different countries. The onset of the epidemic was consistently estimated 2-4 weeks before the reported onset, corresponding to a temporal shift of the reported case incidences in the first wave. In China we discovered multiple peaks that were not observed in the reporting data. For South Korea we identified a second outbreak 4 weeks prior to the increase of reported cases in late August.

Due to the increased use of real-time sequencing, it is envisaged that the method can complement established incidence estimation methods to monitor future pandemics. The method executes within minutes and can be applied to very large data sets.

Words: 297/300 References: [1] Khatri et al., Robust Estimation of Recent Effective Population Size from Number of Independent Origins in Soft Sweeps, Molecular Biology and Evolution (2019), 36, 2040–52 GitHub: https://github.com/trofimovamw/nCovPopDyn

Title:

Stability of SARS-CoV-2 Phylogenies

Authors:

1,2* 3* 1,2* 1,2,4 5 Yatish Turakhia ,​ Nicola De Maio ​, Bryan Thornlow ,​ Landen Gozashti ,​ Robert Lanfear ,​ Conor R. 3,6 ​ 2 ​ 1,2,7​ 8 ​ 9 ​ 3 Walker ,​ Angie S. Hinrichs ,​ Jason D. Fernandes ,​ Rui Borges ,​ Greg Slodkowicz ,​ Lukas Weilguny ,​ David ​ 1,2,7,10 ​ 3,10 ​ 1,2,10 ​ ​ ​ Haussler ,​ Nick Goldman ​ and Russell Corbett-Detig ​ ​ ​

Affiliations:

1. Department of Biomolecular Engineering, University of California Santa Cruz. Santa Cruz, CA 95064, USA 2. Genomics Institute, University of California Santa Cruz, Santa Cruz, CA 95064, USA 3. European Molecular Biology Laboratory, European Bioinformatics Institute (EMBL-EBI), Wellcome Genome Campus, Cambridge CB10 1SD, UK 4. Department of Organismic and Evolutionary Biology and Museum of Comparative Zoology, Harvard University, Cambridge, MA, 02138, USA 5. Department of Ecology and Evolution, Research School of Biology, Australian National University, Canberra, ACT 2601, Australia 6. Department of Genetics, University of Cambridge, Cambridge CB2 3EH, UK 7. Howard Hughes Medical Institute, University of California, Santa Cruz, CA 95064, USA 8. Institut für Populationsgenetik, Vetmeduni Vienna, Wien 1210, Austria 9. MRC Laboratory of Molecular Biology, Cambridge CB2 0QH, UK 10. Correspondence to [email protected], [email protected], [email protected] ​ ​ ​ *Equal contribution

Abstract:

The SARS-CoV-2 pandemic has led to unprecedented, nearly real-time genetic tracing due to the rapid community sequencing response. Researchers immediately leveraged these data to infer the evolutionary relationships among viral samples and to study key biological questions, including whether host viral genome editing and recombination are features of SARS-CoV-2 evolution. This global sequencing effort is inherently decentralized and must rely on data collected by many labs using a wide variety of molecular and bioinformatic techniques. There is thus a strong possibility that systematic errors associated with lab—or protocol—specific ​ ​ ​ ​ practices affect some sequences in the repositories. We find that some recurrent mutations in reported SARS-CoV-2 genome sequences have been observed predominantly or exclusively by single labs, co-localize with commonly used primer binding sites and are more likely to affect the protein-coding sequences than other similarly recurrent mutations. We show that their inclusion can affect phylogenetic inference on scales relevant to local lineage tracing, and make it appear as though there has been an excess of recurrent mutation or recombination among viral lineages. We suggest how samples can be screened and problematic variants removed, and we plan to regularly inform the scientific community with our updated results as more SARS-CoV-2 genome sequences are shared (https://virological.org/t/issues-with-sars-cov-2-sequencing-data/473 and ​ ​ https://virological.org/t/masking-strategies-for-sars-cov-2-alignments/480). We also develop tools for comparing ​ and visualizing differences among very large phylogenies and we show that consistent clade- and tree-based comparisons can be made between phylogenies produced by different groups. These will facilitate evolutionary inferences and comparisons among phylogenies produced for a wide array of purposes. Building on the SARS-CoV-2 Genome Browser at UCSC, we present a toolkit to compare, analyze and combine SARS-CoV-2 phylogenies, find and remove potential sequencing errors and establish a widely shared, stable clade structure for a more accurate scientific inference and discourse. IDENTIFICATION OF SARS-COV-2 RECOMBINANT GENOMES *David VanInsberghe1, Andrew S. Neish1, Anice C. Lowen2,3, Katia Koelle3,4

1Department of Pathology, Emory University, Atlanta, GA, USA 2Department of Microbiology and Immunology, Emory University, Atlanta, GA, USA 3Emory-UGA Center of Excellence for Influenza Research and Surveillance (CEIRS), Atlanta GA, USA 4Department of Biology, Emory University, Atlanta, GA, USA * Correspondence for presenter: [email protected]

Viral recombination can generate novel genotypes with unique phenotypic characteristics, including transmissibility and virulence. Although the capacity for recombination among Betacoronaviruses is well documented, there is limited evidence of recombination between SARS-CoV-2 strains. By identifying the mutations that primarily determine SARS-CoV-2 clade structure, we developed a new approach for detecting and validating recombinant genotypes. Among 68000 SARS-CoV-2 genomes, we detected five putative recombinants that contain multiple phylogenetic markers from distinct clades. The predicted parent clades of these genomes were, with one exception, co-circulating in the country of infection prior to collecting each sample. Our results indicate recombination within SARS-CoV-2 is occurring, but is either not widespread or remains widely undetectable given current levels of genetic diversity. Efforts to monitor the emergence of recombinant genomes should therefore be sustained.

OPTIMIZING VIRAL GENOME SUBSAMPLING BY GENOMIC DIVERSITY AND TEMPORAL DISTRIBUTION

Simone Marini, Carla Mavian, Marco Salemi, Brittany Rife Magalis

Phylogenetic tracing is a method based on viral genome sequencing and evolution used to link infected individuals when transmission history is unknown. By integrating spatial and temporal information in phylogenetic tracing, it is possible to expand the analysis to include phylodynamic inference. Tools that incorporate this expanded approach, such as NextStrain, have been utilized to monitor SARS-CoV-2 evolution and dynamics based on real-time deposition of sequences in databases such as GISAID. Not unlike traditional epidemiological analysis, however, these methods can be affected by sampling bias, which we now know is characteristic of existing SARS-CoV-2 datasets. Not only do the quality and quantity of sequences vary per country, but even regional sample collection policies have varied inconsistently over time and rarely sampled randomly from a representative, stratified population. Both spatial and temporal sampling bias can produce unreliable phylodynamic inferences. Subsampling from the full dataset can overcome sampling bias and is even necessary for analysis of large datasets, such as that of SARS-CoV-2, using sophisticated software (e.g., BEAST) that are not designed to handle thousands of sequences. Whereas subsampling so as to maximize genetic diversity of subpopulations (e.g., individual country) is available, we show that this can lead to skewed temporal sampling distributions. To address this problem, we designed a machine learning approach to optimize genome subsampling according to both genomic diversity and temporal sampling distributions of user-defined subpopulations. The implemented genetic algorithm optimizes these criteria given a user-defined number of sequences per subpopulation (e.g., uniform or proportional to number of cases) with sampling time information and a genomic distance matrix (can also be calculated using the default Jukes-Cantor substitution model). A subpopulation is considered temporally optimized when subsampling approaches a uniform temporal distribution. Our method is available at https://github.com/smarini/tardis-phylogenetics.

Ultrafast Sample Placement on Existing Trees (UShER) Enpowers Real-Time Phylogenetics for the SARS-CoV-2 Pandemic

Abstract: As the SARS-CoV-2 virus spreads through human populations, modern sequencing ​ technologies enable virtually immediate tracing of its evolutionary history and its transmission dynamics. The resulting unprecedented accumulation of viral genome sequences has been essential for epidemiological analyses and is ushering a new era of “genomic contact tracing” – that is, using viral genome sequences to inform on transmission dynamics among human hosts. However, the vast and rapidly growing genome sequence datasets that are becoming available also impose important challenges for efficient and accurate interpretation. In particular, because the viral phylogeny is already so large – and will undoubtedly grow many fold – placing new sequences onto the tree has emerged as an important challenge for real-time genomic contact tracing. Additionally, visualization tools for genomic datasets on this scale are woefully inefficient. Here, we resolve these two interrelated issues by implementing an efficient, tree-based data structure encoding the inferred evolutionary history of the virus. We demonstrate that this data structure, as well as other optimizations, can improve the speed of phylogenetic placement of new samples and data visualization by orders of magnitude and make it possible to complete the placements under real-time constraints. Our method can also accurately reconstruct subtrees when added sequences are closely related. We make these tools available to the research community through the UCSC SARS-CoV-2 Genome Browser to enable rapid cross-referencing these results with a myriad of other datasets. The methods described here will enable real time monitoring of viral genome transmission and evolution.

VIRALMSA: MASSIVELY SCALABLE REFERENCE-GUIDED MULTIPLE SEQUENCE ALIGNMENT OF VIRAL GENOMES *Niema Moshiri1 1 ​ D​ epartment of Computer Science and Engineering, UC San Diego, La Jolla, CA, USA *[email protected]

As the COVID-19 pandemic spreads rapidly around the world, public health officials need to be able to answer questions such as "How is COVID-19 spreading through the population?" and "How many individual outbreaks exist within a given community?". With increasing access to sequencing technologies, scientists can analyze the genome sequences of collected SARS-CoV-2 viral samples in order to gain information about to aid in the development of vaccines and drugs as well as to infer the most likely evolutionary history of the virus, which can help epidemiologists track the spread of the virus across populations.

In molecular epidemiology, the identification of clusters of transmissions typically requires the alignment of viral genomic sequence data. However, the computational problem of Multiple Sequence Alignment (MSA) is NP-Hard, and even the fastest of reasonably-accurate heuristics scale poorly with respect to the number of sequences (typically quadratically). The epidemiological use of the evolutionary history of the virus is primarily useful if it can be updated in real-time, but the computational complexity of MSA coupled with the exponentially-growing number of available SARS-CoV-2 genomes renders such real-time analysis infeasible.

When a high-confidence viral reference genome is available (as is the case with SARS-CoV-2), MSA can be algorithmically accelerated by performing pairwise alignments between each sequenced genome and the reference genome, and then using the positions of the reference genome as anchors with which to merge the individual alignments into a single MSA.

We introduce ViralMSA, a user-friendly reference-guided multiple sequence alignment tool that was built to enable the alignment of ultra-large viral genome datasets. It scales linearly with the number of sequences, and it is able to align tens of thousands of full viral genomes in seconds. ViralMSA is freely available at https://github.com/niemasd/ViralMSA as an open-source software ​ ​ project. THE IMPACT OF SARS-COV-2 VIRAL DYNAMICS ON POPULATION LEVEL SPREAD, SUPER-SPREADER EVENTS, AND OPTIMIZATION OF MASKING AND ANTIVIRAL THERAPY

*Ashish Goyal

Abstract: Approximately 10 months into the pandemic, the world is still struggling to contain the spread of SARS-CoV-2 and to limit the severity of cases. More than 35 million cases and over 1 million deaths have occurred globally with continued severe impact on the world economy. In this talk, I will describe mathematical models based on the observed viral shedding patterns of infected people. These models are intended to optimize prevention and treatment strategies for the virus. I will demonstrate that targeting super-spreader events would have a disproportionate impact on lowering cases, that slight increases in mask uptake or efficacy would curb exponential growth of cases, and that early treatment could be vital to successful treatment of the virus.

KINETICS OF SARS-COV-2 INFECTION IN THE HUMAN UPPER AND LOWER RESPIRATORY TRACTS AND THEIR RELATIONSHIP WITH INFECTIOUSNESS

Ruian Ke1, Carolin Zitzmann1, Ruy M. Ribeiro1, Alan S. Perelson1. 1T-6 Theoretical Biology and Biophysics, Theoretical Division, Los Alamos National Laboratory, NM87545, USA.

SARS-CoV-2 is a human pathogen that causes infection in both the upper respiratory tract (URT) and the lower respiratory tract (LRT). The viral kinetics of SARS-CoV-2 infection and how they relate to infectiousness and disease progression are not well understood. Here, we develop viral dynamic models of SARS-CoV-2 infection in both the URT and LRT. Fitting the models to a set of published viral load data from patients with likely infection dates known, we estimated that infected individuals with a longer incubation period had lower rates of viral growth, took a longer time to reach peak viremia in the URT, and had higher chances of presymptomatic transmission. We then developed a model linking viral load to infectiousness. We found that to explain the substantial fraction of transmissions occurring presymptomatically, a person’s infectiousness should depend on a saturating function of the viral load, making the logarithm of the URT viral load a good surrogate of infectiousness. Comparing the roles of target-cell limitation, the innate immune response, proliferation of target cells and spatial infection in the LRT, we found that spatial dissemination in the lungs is likely to be an important process in sustaining the prolonged high viral loads. Overall, our models provide a quantitative framework for predicting how SARS- CoV-2 within-host dynamics determine infectiousness and represent a step forward towards quantifying how viral load dynamics and the immune responses determine disease severity.

MODELING SARS-COV-2 VIRAL KINETICS AND ASSOCIATION WITH MORTALITY IN HOSPITALIZED PATIENTS RESULTS FROM THE FRENCH COVID-19

*Guillaume Lingas

The characterization of SARS-CoV-2 viral kinetics in hospitalized patients and its association with mortality is controversial. We analyzed death and nasopharyngeal viral kinetics in 655 hospitalized patients from the prospective French-Covid19 cohort. The model predicted a median (Inter Quartiles Range) infection time occurring 4.8 (4.7-5.0) days before symptom onset, with a peak viral load at 1.1 (-2.1-1.6) day before symptom onset. After peak, viral load declined more slowly in in patients with age ≥ 65 (p <10-4) leading to a delayed viral clearance occurring 18 (13.3-22.5) days after symptom onset as compared to 14 (10.7-16.2) days in younger patients (p <10-3). In multivariate analysis, the risk factors associated with mortality were age ≥ 65, male gender and presence of chronic pulmonary disease (HR >2.0). Using a joint model, viral load was an independent predictor of the risk of death (HR=1.3, p <10-3). Finally, we used our model to simulate the effects of antiviral treatment on viral dynamics and survival. Initiating an antiviral treatment upon hospital admission, we predict that a treatment able to reduce viral production by 90% would shorten the time to viral clearance by 3 and 5 days in patients of age < 65 and ≥ 65, respectively. The effects on mortality would be particularly sensible in the latter population, with a reduction of mortality of 4 to 7% depending on the presence of comorbidities. This suggest that early administration of effective antiviral treatments could reduce viral load and hence mortality in most at risk’s patients.

RKI_COVIDTestCalculator: A standalone GUI to assess testing- and quarantine strategies for incoming travelers, contact person management and de-isolation

Wiep van der Toorn1,2, Djin-Ye Oh3, *Max von Kleist1,2,4, on behalf of the working group on SARS-CoV Diagnostics at RKI

1,2 Systems Medicine of Infectious Disease (P5) and Bioinformatics (MF1), Methodology and Research Infrastructure, Robert Koch Institute Berlin, Germany 3 FG17 Influenza and other respiratory viruses, Department of infectious disease, Robert Koch Institute Berlin, Germany 4 German COVID Omics Initiative (deCOI)

In 2019/2020 SARS-CoV2 turned into a global pandemic. Non-pharmaceutical interventions including quarantine of exposed individuals and isolation of infected individuals are key to controlling the epidemic spread of the virus and have led to marked declines in infection and fatalities in many countries. In addition, testing strategies can help to identify infected individuals and tailor quarantine measures. In August, Germany introduced voluntary SARS- CoV2 testing for all incoming travelers from high risk areas as a possibility to circumvent- or shorten quarantine and to cushion the societal burden of the pandemic.

We built a user-friendly standalone GUI that calculates residual risk (probability of being infectious upon release from quarantine or isolation), risk reduction and test efficacy for arbitrary testing strategies. The user determines whether the time of exposure/infection (or symptom onset) is known, the intended duration of isolation/quarantine and the time points for SARS-CoV2 PCR tests. Underneath, we implemented the analytical solution of a stochastic transit compartment model of the infection time course that captures published temporal changes and variabilities in test sensitivities, incubation- and infectious periods, as well as times to symptom onset.

Using the tool, we estimated that a SARS-CoV2 screening of travelers at the point of entry, with an unknown time of exposure, reduces the risk about 7.6 fold when combined with symptom screening and 3.5 fold otherwise. By contrast, a 7days quarantine combined with a single test on day 5 would reduce the risk about 26-fold (7 fold without symptom screening). For contact person management (known time of exposure) a 5 days quarantine with subsequent test reduces the risk about 15-fold. We predicted that de-isolation of infected individuals 9 days after symptom onset or 13 days after exposure reduces the risk by >99%.

This tool will be further extended to include antigen- and point-of-care tests and is freely available under https://github.com/Wvandertoorn/CovidStrategyCalculator

A PROTOTYPE QSP MODEL OF THE IMMUNE RESPONSE TO SARS-CoV-2 FOR COMMUNITY DEVELOPMENT

Wei Dai†*, Rohit Rao†*, Anna Sher, Nessy Tania, CJ Musante, Richard Allen

*Presenting authors †Joint first authors

Early Clinical Development, Pfizer Worldwide Research, Development and Medical, Cambridge, MA, USA

The application of model-based approaches, including quantitative systems pharmacology models (QSP), have accelerated the development of some novel therapeutics. While the rationale for development of a QSP model of COVID-19 is clear, the development of disease- scale mechanistic models can be a slow process, often taking years to be validated and considered mature. Furthermore, emerging data may make any disease-scale model quickly obsolete. Given that following a traditional QSP model development timeline would not be feasible in order to have an impact, we aim to quickly publish a prototype model. It is our hope that releasing a prototype model at this relatively early stage will encourage community engagement and ultimately accelerate model refinement and testing through an open-science approach. The model accounts for the influence of key mediators relevant to COVID-19 pathophysiology including, interactions between viral dynamics, the major host immune response mediators, and alveolar tissue damage and regeneration. The model is able to qualitatively capture two physiologically relevant outcomes following infection: a “healthy” immune response that is appropriately activated and resolves after adequate clearance of the virus and a pathophysiological uncontrolled inflammatory response that is characteristic of severe cases of COVID-19 presenting with ARDS. Our simulations also suggest a mechanistic basis for how both a dampened as well as overly reactive immune response might lead to pathological outcomes with foreseeable implications for therapeutic intervention. The model further provides a framework for numerous potential adaptations and extensions including the development of virtual populations to study disease heterogeneity and relatively straightforward incorporation of anti-viral and immunomodulatory treatments.

Correspondence: Richard Allen, [email protected]

A COALITION-DRIVEN EFFORT TO MODEL SARS-COV-2 SPREAD AND IMMUNE RESPONSE IN TISSUE

*Paul Macklin, on behalf of the COVID-19 Modeling Coalition

The 2019 novel coronavirus, SARS-CoV-2, is a pathogen of critical significance to international public health. Knowledge of the interplay between molecular-, cellular-, and multicellular-scale processes that drive disease dynamics is limited. Multiscale simulation models can shed light on these dynamics, identify actionable “choke points” for intervention, screen potential therapies, and identify potential biomarkers that differentiate patient outcomes. In this talk, we present progress by a multi-institution, multi-disciplinary coalition of over 40 mathematical biologists, immunologists, virologists, pharmacologists, and others to build a comprehensive multiscale model of SARS-CoV-2 infection dynamics and immune response. The team has adopted open science approaches to accelerate scientific communication, including full open source release of the simulation source code, frequent updates to an open access preprint, and the use of free, cloud-hosted models that can be run without downloading and compiling computer code.

We will demonstrate and explore the current spatiotemporal agent-based model prototype, which includes: intracellular virus and chemokine transport, virus-ACE2 receptor binding, receptor trafficking, viral replication dynamics (and subsequent viral release), infected cell phenotypic responses (including pyroptotic death and interferon signaling), cell-cell communication, a broad variety of macrophage behaviors, trafficking of antigen-presenting cells, systemic immune expansion, immune cell recruitment, T cell and macrophage attacks on infected cells, and phagocytosis.

This coalition-based approach is developing model components in parallel and coordination, allowing us to rapidly advance towards a framework that can drive many independent investigations on COVID-19, and can be generalized and adapted to other infectious diseases. Moreover, the novel mix of domain experts is fueling creative advances and new technical capabilities beyond infectious diseases. We anticipate that this progress will drive advances in immunology, inflammation, CAR T cell therapies, and virus-driven carcinogenesis for years to come. Interested members of the audience can try this open source framework live in a web browser at https://nanohub.org/tools/pc4covid19.

Systemic and tissue-level quantitative modelling distinguishes the immune response to SARS-CoV-2

Adrianne Jenner1,2, Vivienne Crowe3, Sofia Alfonso4, Rosemary Aogo5, Penelope A. Morel6, Courtney L. Davis7*, Amber M. Smith5*, Morgan Craig1,2, 4* 1CHU Sainte-Justine Research Centre, Canada 2Department of Mathematics and Statistics, Université de Montréal, Canada 3Department of Mathematics and Statistics, Concordia University, Canada 4Department of Physiology, McGill University, Canada 5Department of Pediatrics, University of Tennessee Health Science Center, USA 6Department of Immunology, University of Pittsburgh, Pittsburgh, USA 7Natural Science Division, Pepperdine University, USA *Co-senior authors The primary distinction between severe and mild COVID-19 infections is the immune response. Disease severity and fatality has been observed to correlate with lymphopenia (low blood lymphocyte count) and increased levels of inflammatory cytokines and IL-6 (cytokine storm), damaging dysregulated macrophage responses, and T cell exhaustion due to limited recruitment. The exact mechanism driving the dynamics that ultimately result in severe COVID-19 manifestation remain unclear. To delineate mechanisms regulating differential immune responses to SARS-CoV-2, we have developed tissue- and systemic-level models of the immune response to infection with the goal of pinpointing what may be causing dysregulated immune dynamics in severe cases. At the tissue level, we been working as part of the international SARS-CoV-2 Tissue Simulation Coalition (physicell.org/covid19) to build a computational framework to study SARS-CoV-2 in the tissues based on an open-source computational cell-based software that combines an agent-based modelling framework with partial differential equation diffusion models. This model includes a cell-intrinsic ODE system to model endocytosis, viral replication and pyroptosis, combined with force equations governing cell motility and built in cell-subtract uptake and secretion models to account for virus, chemokine and pro-inflammatory cytokine dynamics. With this platform, we have investigated how the level of pro-inflammatory cytokines influence immune cell recruitment into the infected tissue and how this correlates with tissue damage. In parallel, we constructed a systemic, within-host mechanistic mathematical model linking viral kinetics to the innate and adaptive immune response. This model accounts for the interactions between viral load, viral strain, infected and damaged epithelial tissue cells in the lungs, immune cell subsets (primarily tissue-resident and inflammatory macrophages, CD8 T cells, monocytes, and neutrophils), and inflammatory cytokines (i.e. IFN-1, GM-CSF, G-CSF, IL-6), and recapitulates mild and severe COVID-19 presentations. Through sensitivity analysis and the expansion of virtual patient cohorts, we analyzed the mechanisms regulating the diversity of immune responses to SARS-CoV-2 to identify those that predispose individuals to particularly to severe disease. Together, these platforms represent a comprehensive framework that will improve our understanding of SARS-CoV-2 infection dynamics and immune response. POSTERS

1. CAPTURING 35 NATIONAL SCHOOL CLOSURE INTERVENTIONS IN A MODEL OF COVID-19 DIAGNOSES AND DEATHS, *R. Belew

2. HIGH-DENSITY AMPLICON SEQUENCING IDENTIFIES COMMUNITY SPREAD AND ONGOING EVOLUTION OF SARS-COV-2 IN THE SOUTHERN UNITED STATES, *J. Landis

3. HOSPITAL CAPACITY AND COVID-19 MORTALITY, *M. Alkuzweny

4. APPROPRIATE APPROACHES TO ESTIMATE THE EARLY EPIDEMIC GROWTH RATE AND THE REPRODUCTIVE NUMBER R0 OF SARS-COV-2 AND IMPLICATIONS FOR VACCINATION, *R. Ke

5. SARS-COV-2 GENOMIC AND QUASISPECIES ANALYSES IN CANCER PATIENTS REVEAL RELAXED INTRAHOST VIRUS EVOLUTION, *L. Goes

6. EFFECTIVENESS OF MASSIVE TRAVEL RESTRICTIONS ON MITIGATING OUTBREAKS OF COVID-19 IN CHINA, *X. Chen

7. LINEAGE-LEVEL DIVERSIFICATION RATES REFLECT LONGITUDINAL CHANGES IN COVID-19 EPIDEMIC CHARACTERISTICS, *R. Miller

8. VARIABLE ROUTES TO GENOMIC AND HOST ADAPTATION AMONG CORONAVIRUSES, *V. Montoya

9. PHYLODYNAMIC ANALYSES SUPPORT THE ASSOCIATION OF THE INITIAL EXPANSION OF A LARGE SARS-COV-2 LINEAGE (A.2) TO A FUNERAL IN NORTHERN SPAIN, *M. Thomson

10. Withdrawn

11. WRONG PERSON, PLACE, AND TIME: VIRAL LOAD AND CONTACT NETWORK STRUCTURE PREDICT SARS-COV-2 TRANSMISSION AND SUPER-SPREADING EVENTS, *B. Mayer

12. ESTIMATING THE STATE OF THE COVID-19 EPIDEMIC IN FRANCE USING A NON-MARKOVIAN MODEL, *R. Forien

13. Withdrawn

14. MASK MEDIATED REDUCTION OF EXPOSURE VIRAL LOAD MAY RESULT IN REDUCTION OF SARS-COV-2 TRANSMISSION, *F. Cardozo-Ojeda

15. A NETWORK SEIR MODEL: QUARANTINE EFFECTS AND GENETIC EVOLUTION, *V. Marquioni

16. PROCESS MEMORY IS KEY TO CAPTURING COVID-19 EPIDEMIOLOGICAL DYNAMICS, *S. Alizon

17. CONTROLLING COVID-19 VIA TEST-TRACE-QUARANTINE, *C. Kerr POSTERS

18. Withdrawn

19. ESTIMATION OF SARS-COV-2 MORTALITY DURING THE EARLY STAGES OF AN EPIDEMIC: A MODELING STUDY IN HUBEI, CHINA, AND SIX REGIONS IN EUROPE, *A. Hauser

20. HOSPITAL DEMAND UNDER INTERVENTION DURING THE COVID-19 PANDEMIC : A MODELING STUDY, *K. Hayashi

21. TIME TREND OF CASE FATALITY RISK OF COVID-19 IN JAPAN, *T. Kayano

22. PHENOMENOLOGICAL AND MECHANISTIC MODEL COMPARISON FOR THE FORECAST OF AN EPIDEMIC USING COVID-19 REPORTED DATA IN CHINA, *T. Miyama

23. GLOBAL GEOGRAPHIC AND TEMPORAL ANALYSIS OF SARS-COV-2 HAPLOTYPES NORMALIZED BY COVID-19 CASES, *S. Justo Arevalo

24. SARS-COV-2 ON CAMPUS, MODELING THE EDGE OF FAILURE UNDER ASSUMPTIONS OF CONTACT NUMBER AND COMPLIANCE PARAMETERS, *G. Corey

25. THE UNINTENDED CONSEQUENCES OF INCONSISTENT PANDEMIC CONTROL POLICIES, *S. Scarpino

26. VARIATION IN SARS-COV-2 FREE-LIVING SURVIVAL AND ENVIRONMENTAL TRANSMISSION CAN MODULATE THE INTENSITY OF EMERGING OUTBREAKS, *S. Scarpino

27. ASSESSING LINKED SELECTION AND LONG-DISTANCE ASSOCIATIONS OF FUNCTIONAL MUTATIONS IN SARS-COV2 VARIANTS IN INDIA, *I. Singh Rawal

28. CROWDING AND THE SHAPE OF COVID-19 EPIDEMICS, *S. Scarpino

29. APPLYING V-PIPE TO SARS-COV-2, *I. Topolsky

30. RECONSTRUCTING THE EARLY GLOBAL DYNAMICS OF UNDER-ASCERTAINED COVID-19 CASES AND INFECTIONS, *T. Russel

31. EPIDEMIOLOGICAL MODELING OF CONFID-19 DEATH RATE EXPLAINS PERVERSE MOBILITY- MORTALITY CORRELATION, *D. Lewis

32. GENOMIC EPIDEMIOLOGY OF THE EARLY STAGES OF SARS-COV-2 OUTBREAK IN RUSSIA, *G. Bazykin

33. DIFFERING IMPACTS OF GLOBAL AND REGIONAL RESPONSES ON SARS-COV-2 TRANSMISSION CLUSTER DYNAMICS, *B. Rife Magalis POSTERS

34. FEASIBILITY OF CONTAINING COVID-19 BY SYMPTOM BASED CASE ISOLATION, * R. Kinoshita

35. Withdrawn

36. OPTIMISING TIME-LIMITED NON-PHARMACEUTICAL INTERVENTIONS FOR COVID-19 OUTBREAK CONTROL, *A. Morgan

CAPTURING 35 NATIONAL SCHOOL CLOSURE INTERVENTIONS IN A MODEL OF¬ COVID-19 DIAGNOSES AND DEATHS¬ ¬ * Richard K. Belew, [email protected]¬ ¬ Cliff Kerr^+¬ Jasmina Panovska-Griffiths^#¬ Dina Mistry^+¬ ¬ * Univ. California, San Diego¬ ¬ ^+ Institute for Disease Modeling, Bellevue, WA¬ ^# University College London¬ ¬ COVID-19 continues to spread around the world and modelling plays an¬ important role in informing policy. An individual-based model called¬ Covasim has recently been fit to data regarding confirmed cases and¬ deaths experience in the United Kingdom during the first half of 2020,¬ and then used it to evaluate alternative intervention strategies there¬ [#JPG20]. We extend this methodology to consider data from 35 other¬ countries, and use a database of international intervention specifics¬ called the COVID-19 CONTROL STRATEGIES LIST to retrospectively model¬ interventions employed in these countries. Because the age¬ distribution of populations is a key feature of the COVID-19 pandemic¬ and contacts among young people are often age-stratified and may play¬ an especially important role, we focus here on school closure¬ interventions.¬ ¬ Individual countries varied considerably in both the dates on which¬ they imposed school closings, and in the levels (kindergarten,¬ primary, secondary, university) specified. Critically, the¬ age-stratified sub-populations supported by Covasim allow fine-grained¬ specification of just which individuals are affected by school¬ closures at each educational level. Across the 35 countries the model¬ was calibrated to the country's epidemic, simulations were first run¬ without the intervention, and data on confirmed cases and deaths was¬ used to fit key Covasim parameters. Next, specific intervention¬ strategies employed by each country were converted into specifications¬ for Covasim, and the same simulation parameters fit to a second model¬ with the interventions considered. In the 10 countries where there was¬ a significant difference between models, those incorporating school¬ closures were considerably better fits than those without. Since both¬ models' parameters are optimized and evaluated using the same¬ criterion, improved fit with the intervention model may be taken as¬ evidence that the modelled interventions were useful, at least in these¬ 10 countries, in describing observed data.¬ ¬ [300 words]¬ ¬ JPG20: Panovska-Griffiths, J. et al, The Lancet Child Adolescent¬ Health, 2020/08/11, DOI:10.1016/S2352-4642(20)30250-9¬ ¬ Covasim: https://github.com/InstituteforDiseaseModeling/covasim¬ ¬ Covid-19 Control Strategies List:¬ https://github.com/amel-github/covid19-interventionmeasures¬ HIGH-DENSITY AMPLICON SEQUENCING IDENTIFIES COMMUNITY SPREAD AND ONGOING EVOLUTION OF SARS-COV-2 IN THE SOUTHERN UNITED STATES

*Justin T. Landis (1,2), Ryan P. McNamara (1,2), Carolina Caro-Vegas (1,2), Razia Moorad (1,2), Linda J. Pluta (1,2), Anthony B. Eason (1,2), Cecilia Thompson (3, 4), Aubrey Bailey (5), Femi Cleola S. Villamor (1,2), Philip T. Lange (1,2), Jason P. Wong (1,2), Tischan Seltzer (1,2), Jedediah Seltzer (1,2), Yijun Zhou (1,2), Wolfgang Vahrson (6), Angelica Juarez (1,2), James O. Meyo (1,7) , Tiphaine Calabre (1,8), Grant Broussard (1,7), Ricardo Rivera-Soto (1,7), Danielle L. Chappell (1,9), Ralph S. Baric (2,10), Blossom Damania (1,2), Melissa B. Miller (3, 4), Dirk P. Dittmer (1,2)

The University of North Carolina at Chapel Hill School of Medicine, Department of Microbiology and Immunology, (1), Lineberger Comprehensive Cancer Center (2), Department of Pathology and Laboratory Medicine (3), UNC Medical Center, Clinical Microbiology Laboratory (4), Kuopio Center for Gene and Cell Therapy, Finland (5), Basel, Switzerland (6), Genetics Curriculum (7), École supérieure de Chimie Physique Électronique (CPE) Lyon, France (8), Department of Pharmacology (9), Department of Epidemiology (10)

SARS-CoV-2 is constantly evolving. Prior studies have focused on high case-density locations, such as the Northern and Western metropolitan areas in the U.S. This study demonstrates continued SARS-CoV-2 evolution in a suburban Southern U.S. region by high-density amplicon sequencing of symptomatic cases. 57% of strains carried the spike D614G variant. The presence of D614G was associated with a higher genome copy number and its prevalence expanded with time. Four strains carried a deletion in a predicted stem loop of the 3’ untranslated region. The data are consistent with community spread within the local population and the larger continental U.S. No strain had mutations in the target sites used in common diagnostic assays. The data instill confidence in the sensitivity of current tests and validate “testing by sequencing” as a new option to uncover cases, particularly those not conforming to the standard clinical presentation of COVID-19. This study contributes to the understanding of COVID-19 by providing an extensive set of genomes from a non-urban setting and further informs vaccine design by defining D614G as a dominant and emergent SARS-CoV-2 isolate in the U.S.

HOSPITAL CAPACITY AND COVID-19 MORTALITY

1 2 3 *Manar Alkuzweny ,​ Anita Raj PhD ,​ Sanjay Mehta MD ​ ​ ​ 1. Department of Family Medicine and Public Health, University of California San Diego, 9500 Gilman Drive, La Jolla, CA, 92093 2. Department of Medicine, University of California San Diego, 9500 Gilman Drive, La Jolla, CA, 92093 3. Departments of Medicine and Pathology, University of California San Diego, 9500 Gilman Drive, La Jolla, CA, 92093 and Department of Medicine, San Diego Veterans Affairs Medical Center, 3350 Via La Jolla Drive, San Diego, CA 92161

Presenting author: Manar Alkuzweny 4919 Via Cinta San Diego, CA 92122 Email: [email protected] Phone: 858-945-5419

Background: The US bears the highest caseload and deaths due to COVID-19 in the world. Adults over the age of 65 are disproportionately more likely to progress to severe disease and experience mortality. In addition, the case fatality rate (CFR) of patients admitted to the ICU is >25%. Given variations in hospital capacity and the population density of the elderly across the US, it is important to understand which regions will experience disproportionate burdens of COVID-19.

Methods: We examined the relationship between hospital beds and ICU beds per 1000 persons over the age of 65 and CFR using Spearman’s rho. We also mapped CFR for all US counties to understand which counties are experiencing a severe burden of COVID-19, as well as hospital and ICU beds per 1000 persons over the age of 65 to understand which counties have limited capacity relative to their elderly population.

Results: We found a significant negative association between hospital beds per 1000 persons over the age of 65 (p<0.0001) and a significant positive association between ICU beds per 1000 persons over the age of 65 (p<0.0001).

Discussion: The relationship between hospital beds per 1000 persons over the age of 65 and CFR indicates that limited access to inpatient care increases the likelihood of mortality for older COVID-19 patients, while the positive association between ICU beds per 1000 persons over the age of 65 and CFR may be due to disparities in ICU capacity across rural and urban counties. Critically ill patients in rural hospitals are likely referred to urban hospitals with more resources, leading to higher mortality in urban counties with higher ICU bed capacity. The variability in hospital and ICU bed capacity suggests state resources will need to be strategically allocated to prepare for a second wave of COVID-19. APPROPRIATE APPROACHES TO ESTIMATE THE EARLY EPIDEMIC GROWTH RATE AND THE REPRODUCTIVE NUMBER R0 OF SARS-COV-2 AND IMPLICATIONS FOR VACCINATION

*Ruian Ke

Affiliation: T-6 Theoretical Biology and Biophysics, Theoretical Division, Los Alamos National Laboratory, NM87545, USA.

Abstract: The early epidemic growth rate and the reproductive number R0 for SARS-CoV-2 are two fundamental epidemiological parameters for the understanding of COVID-19 transmission dynamics, evaluation of control strategies, and prediction of the level of herd immunity needed to stop transmission. Yet their values are highly debated in the literature, partly due to rapidly evolving knowledge of key parameter values for this diseasee, such as serial intervals. I will discuss appropriate approaches to accurately estimate the two fundamental epidemiological parameters for a novel infectious disease outbreak, such as COVID-19. Since early COVID-19 outbreak in January, we worked on estimating these quantities. Our work represents one of the first to conclude that population wide social distancing efforts are necessary to stop virus transmission (in early February). We argue that existing evidence suggests a highly infectious virus high values of R0 (likely between 4.0 and 7.1 in highly populated areas across China, Europe and the US). This translates to high herd immunity thresholds. We further analyze how vaccination schedules depends on R0, the duration of vaccine-induced immunity to SARS-CoV-2, and show the importance of measuring individual-level heterogeneity in vaccine induced immunity to predict how long a herd immunity can be maintained.

SARS-COV-2 GENOMIC AND QUASISPECIES ANALYSES IN CANCER PATIENTS REVEAL RELAXED INTRAHOST VIRUS EVOLUTION Juliana D. Siqueira1, Livia R. Goes1,2, Brunna M. Alves1, Pedro S. de Carvalho1, Claudia Cicala2, James Arthos2, João P.B. Viola4, Andréia C. de Melo3 and *Marcelo A. Soares1,5 1Programa de Oncovirologia, Instituto Nacional de Câncer. Rio de Janeiro, RJ, Brazil, 2Laboratory of Immunoregulation, National Institute of Allergy and Infectious Diseases, Bethesda, MD, USA, 3Divisão de Pesquisa Clínica e Desenvolvimento Tecnológico, Instituto Nacional de Câncer, Rio de Janeiro, RJ, Brazil, 4Programa de Imunologia e Biologia de Tumores, Instituto Nacional de Câncer, Rio de Janeiro, RJ, Brazil, 5Departamento de Genética, Universidade Federal do Rio de Janeiro, Rio de Janeiro, RJ, Brazil

Cancer patients are more prone to clinically evolve to more severe COVID-19 conditions, but the determinants of such a more severe outcome remain unknown. In the present, study we have determined the full-length SARS-CoV-2 genomic sequences of cancer patients and healthcare workers (HCW, non-cancer controls) by deep sequencing and investigated the within-host viral quasispecies of each infection, quantifying intrahost genetic diversity.

SARS-CoV-2+ swabs from 57 cancer patients and 14 HCW from the Brazilian National Cancer Institute were collected in April–May 2020. Complete genome was amplified using ARTIC network V3 multiplex primers followed by next-generation sequencing. Consensus sequences were assembled and extracted and intrahost single nucleotide variants were identified. Maximum likelihood phylogenetic analysis was performed using PhyMLv.3.0 and lineages were classified using Pangolin and CoV-GLUE.

Phylogenetic analysis showed that all but one strain belonged to clade B1.1. Four genetically linked mutations known as the globally dominant SARS-CoV-2 haplotype (C241T, C3037T, C14408T and A23403G) were found in the majority of consensus sequences as well as SNV signatures of characterized Brazilian genomes. Cancer patients displayed a significantly higher intrahost viral genetic diversity compared to HCW. Most intrahost SNVs in both groups were related to APOBEC and ADAR activities. Intrahost genetic diversity in cancer patients was independent of SARS-CoV-2 Ct values, and was not associated with disease severity, use of corticosteroids, or use of antivirals, characteristics that could influence viral diversity. However, patients with pulmonary metastases alone or grouped with all patients with metastatic cancer showed a borderline significant trend to higher intrahost diversity when compared to other cancer patients.

Cancer patients carried significantly higher numbers of minor variants compared to non- cancer counterparts, which may impact viral pathogenesis. Further studies on SARS-CoV-2 diversity in especially vulnerable patients will shed light onto the understanding of the basis of COVID-19 different outcomes in humans.

Marcelo Alves Soares Program of Oncovirology - Instituto Nacional de Câncer Rua André Cavalcanti, 37 - 4o andar, 20.231-150 Rio de Janeiro – RJ - Brazil Tel: +55 21 3207-6591, E-mail: [email protected] EFFECTIVENESS OF MASSIVE TRAVEL RESTRICTIONS ON MITIGATING OUTBREAKS OF COVID-19 IN CHINA

∗Xingru Chen∗1 and Feng Fu†1,2

1Department of Mathematics, Dartmouth College, Hanover, NH 03755, USA 2Department of Biomedical Data Science, Geisel School of Medicine at Dartmouth, Lebanon, NH 03756, USA

September 15, 2020

Abstract

In the very early stage of an unprecedented outbreak of COVID-19 started in the epicenter, Wuhan, Hubei Province, China, the Chinese government imposed by far the largest scale of strict travel restrictions on more than 11 million people (beyond) on January 23, 2020, amid the busiest period of the year for domestic travels (chunyun, travels made during the Lunar New Year). Such massive travel restrictions have caused dramatic reduction in travel volume, not only for the outflow from Wuhan (Hubei), but also nationwide. Control measures like this helps reduce the number of imported cases to other provinces, which can possibly slowdown the onset of epidemic outbreaks in other regions and potentially weaken the impact of the disease. Here, we are interested in estimating the effectiveness of such massive travel restrictions in the mitigation of disease impact using a data driven approach.

[email protected][email protected]

1 LINEAGE-LEVEL DIVERSIFICATION RATES REFLECT LONGITUDINAL CHANGES IN COVID-19 EPIDEMIC CHARACTERISTICS

*Rachel L. Miller1,2, Angela McLaughlin1,2, Vincent Montoya1, and Jeffrey B. Joy1,2,3

1British Columbia Centre for Excellence in HIV/AIDS, St. Paul’s Hospital, Vancouver, Canada; 2Bioinformatics, University of British Columbia, Vancouver, Canada; 3Department of Medicine, University of British Columbia, Vancouver, Canada. Presenting author correspondence: BC Centre for Excellence in HIV/AIDS, 608-1081 Burrard St, Vancouver, BC V6Z 1Y6, [email protected]

Background Forces such as transmission and selection influence viral evolution such that epidemic characteristics are reflected in phylogenetic trees, and thus phylodynamic analyses of genomic sequence data can reveal epidemic dynamics. Here, we test the hypothesis that changes in longitudinal lineage-level diversification rates reflect changes in epidemic dynamics.

Methods Sequences from Canada, UK, USA and Australia were obtained from GISAID, aligned separately and used to infer rooted phylogenetic trees. Lineage-level diversification rates were calculated individually for all tips on all days with least 10 sequences were available.

Results The daily-measured maximum diversification rate of SARS-CoV-2 reveals increases ahead of and alongside increases in number of new cases for Australia, UK and Canada. Peaks in change in mean and maximum diversification rates co-occur with peaks in number of new cases for Australia and UK, but occur before Australia’s second and Canada’s first wave. Canadian sequence collection dates are released at month-level precision, and thus all major diversification rate peaks coincide with once-monthly peaks in sequence collection. In the USA, maximum diversification rate demonstrates no peaks, and the change in mean and maximum diversification rates consistently show many peaks with no obvious pattern.

Conclusions Phylogenetically-derived SARS-CoV-2 maximum diversification rates may provide a framework for anticipation of increases in new cases, which may be valuable for effective allocation of public health response resources. Change in mean and maximum diversification rates may have predictive capacity in well-sampled countries (ie. Australia), but decrease in practicality in less comprehensively-sampled countries (ie. USA), where a newly sampled sequence may be genetically distant from its phylogenetic neighbours despite being genetically close to un- sampled circulating genomes. Thus, longitudinal changes in diversification rates may reflect epidemiological characteristics such as sampling proportion and contact tracing effectiveness and additionally demonstrate the potential epidemiological value of achieving high sampling proportion and appropriately precise SARS-CoV-2 data availability.

VARIABLE ROUTES TO GENOMIC AND HOST ADAPTATION AMONG CORONAVIRUSES

Vincent Montoya1*, Gideon J. Mordecai2, Rachel L. Miller1,3, Angela McLaughlin1,3, Jeffrey B. Joy1,2,3

1British Columbia Centre for Excellence in HIV/AIDS, Vancouver, BC

2Department of Medicine, University of British Columbia, Vancouver, BC 3Bioinformatics Programme, University of British Columbia, Vancouver, BC

The role and outcome of natural selection on viral pathogens in different host species results in genomic adaptations that facilitate colonization of novel hosts. Here, we analyze, quantify and compare viral adaptation in genomic sequence data derived from seven zoonotic events in the Coronaviridae family among primary, intermediate and terminal hosts. Rates of nonsynonymous (dN) and synonymous (dS) changes on specific amino acid positions were quantified for each open reading frame (ORF). Purifying selection accounted for 77% of all sites under selection. Diversifying selection was frequently observed in viruses infecting the primary hosts of each virus, and predominantly occurred in the 5 ’ORF1ab genomic region. Within three of the four intermediate hosts, diversifying selection on the spike gene was observed either solitarily or in combination with ORF1ab and other genes. Consistent with previous evidence, pervasive diversifying selection on coronavirus spike genes corroborates the role this protein plays in host cellular entry, adaptation to new hosts, and evading host immune responses. Amongst human coronaviruses, there was a significant inverse correlation between the number of sites under positive selection and the estimated years since the virus was introduced into the human population. Abundant diversifying selection observed in SARS-CoV-2 suggests the virus remains in the adaptive phase of the host switch, typical of recent host switches. A mechanistic understanding of where, when, and how genomic adaptation occurs in coronaviruses following a host shift is crucial for vaccine design, public health responses, and predicting future pandemics. PHYLODYNAMIC ANALYSES SUPPORT THE ASSOCIATION OF THE INITIAL EXPANSION OF A LARGE SARS-COV-2 LINEAGE (A.2) TO A FUNERAL IN NORTHERN SPAIN

María Iglesias-Caballero1, *Michael M Thomson2, Sara Monzón3, Pilar Jiménez4, Sarai Varona3, Isabel Cuesta3, Ángel Zaballos4, Mercedes Jiménez4, Francisco Pozo1, and Inmaculada Casas1.

1Respiratory Virus and Influenza Unit, National Center of Microbiology, National Influenza Center, Instituto de Salud Carlos III, Majadahonda, Madrid, Spain. 2HIV Biology and Variability Unit, National Center of Microbiology, Instituto de Salud Carlos III, Majadahonda, Madrid, Spain. 3Bioinformatics Unit, Instituto de Salud Carlos III, Majadahonda, Madrid, Spain. 4Genomics Unit, Instituto de Salud Carlos III, Majadahonda, Madrid, Spain.

Background: One of the first SARS-CoV-2 outbreaks in Spain was associated to a multitudinous wake and funeral, which took place on February 23 and 24, respectively, in Vitoria, Basque Country, Spain. Here we perform phylogenetic and phylodynamic analyses on SARS-CoV-2 genomes from people who attended the funeral.

Methods: SARS-CoV-2 whole genome sequences were obtained from nasopharyngeal secretions using next-generation sequencing from 28 individuals, living in four Spanish regions, who had attended the funeral at Vitoria and tested PCR-positive for SARS-CoV-2. Lineage classification was done with Pangolin. Maximum likelihood phylogenetic analyses were performed with IQ-Tree. Temporal and geographic origin of lineages were estimated with the Bayesian method implemented in BEAST 1.8.4, with the substitution rate estimated previously with BEAST with 135 SARS-CoV-2 sequences which exhibited a good temporal signal, according to TempEst.

Results: All but one sequences linked to the funeral were of A.2 lineage (which belongs to S/19B clade, according to GISAID/Nextstrain classifications). Phylogenetic analyses including all high-quality S clade SARS-CoV-2 genomes downloaded from GISAID (after removal of identical sequences, which left 19 from the outbreak) showed clustering of sequences from the outbreak with 736 GISAID sequences from 29 countries (327 from Spain), all but two classified in A.2 lineage, according to Pangolin. Phylodynamic analyses, using similar number of sequences from each country and Spanish region, and considering attendance at the Vitoria funeral as a location trait, supported an origin of A.2 lineage in the funeral outbreak (posterior probability=0.62, with Basque Country, PP=0.18, and Valencia, PP=0.09, coming 2nd and 3rd, respectively). Time of the most recent common ancestor of A.2 was estimated around 12 February 2020 (95% HPD, 31 January-22 February).

Conclusions: Phylodynamic analyses support the association of the initial expansion of A.2 lineage with a funeral that took place on 23-24 February 2020 in the city of Vitoria, Spain.

*Presenting author Address: Michael Thomson National Center of Microbiology Instituto de Salud Carlos III Carretera Majadahonda-Pozuelo, Km. 2 28220 Majadahonda, Madrid, Spain E-mail: [email protected]

WRONG PERSON, PLACE, AND TIME: VIRAL LOAD AND CONTACT NETWORK STRUCTURE PREDICT SARS-COV-2 TRANSMISSION AND SUPER-SPREADING EVENTS Ashish Goyal1, Daniel B Reeves1, E. Fabian Cardozo-Ojeda1, Joshua T Schiffer1,2,3*†, Bryan T. Mayer1†* 1 Vaccine and Infectious Diseases Division, Fred Hutchinson Cancer Research Center 2 Department of Medicine, University of Washington, Seattle 3 Clinical Research Division, Fred Hutchinson Cancer Research Center † These authors contributed equally to the work. * presenting author: [email protected]

SARS-CoV-2 is difficult to contain because most transmissions occur during the pre- symptomatic phase of infection. Moreover, in contrast to influenza, while most SARS-CoV-2 infected people do not transmit the virus to anybody, a small percentage secondarily infect large numbers of people. Here, we designed individual-level transmission models of SARS-CoV-2 and influenza which link observed viral shedding patterns with key epidemiologic features of each virus, including distributions of the number of secondary cases attributed to each infected person (individual R0) and the duration between symptom onset in the transmitter and secondarily infected person (serial interval). We identify that people with SARS-CoV-2 or influenza infections are usually contagious for fewer than two days congruent with peak viral load several days after infection, and that transmission is unlikely below a certain viral load. SARS-CoV-2 super-spreader events with over 10 secondary infections occur when an infected person is briefly shedding at a very high viral load and has a high concurrent number of exposed contacts. While viral shedding of SARS-CoV-2 may last 1-2 weeks longer than influenza, we find that differences in viral kinetics do not adequately explain why super spreader events are more likely with SARS-CoV-2. Rather, our model suggests a person infected with SARS-CoV-2 exposes more people within equivalent physical contact networks than a person infected with influenza, likely due to aerosolization of virus. Our results support policies that limit crowd size in indoor spaces and provide viral load benchmarks for infection control and therapeutic interventions intended to prevent secondary transmission.

Estimating the state of the Covid-19 epidemic in France using a non-Markovian model

Rapha¨elForiena, Guodong Pangb, and Etienne´ Pardouxc

aINRAE, BioSP, Centre INRAE PACA, Domaine St-Paul, 84914 Avignon Cedex FRANCE, [email protected] bThe Harold and Inge Marcus Department of Industrial and Manufacturing Engineering, College of Engineering, Pennsylvania State University, University Park, PA 16802 USA, [email protected] cAix–Marseille Universit´e,CNRS, Centrale Marseille, I2M, UMR 7373 13453 Marseille, France, [email protected]

Covid-19 Dynamics and Evolution October 19th-20th 2020

Abstract At the beginning of an epidemic outbreak, the actual number of infected individu- als is rarely directly observed, and model predictions are used to estimate the current state and future evolution of the epidemic. A large majority of these epidemic models assume that the time spent by infected individuals in various compartments (exposed, infectious, removed) is drawn at random from an exponential distribution with some fixed parameters. Despite the fact that the exponential distribution seldom reflects the true distribution of the exposed and infectious periods, these so-called Markov models have a long history of applications to epidemics, mainly because the large pop- ulation limit of these models follows a simple system of ordinary differential equations. Recently, Pang and Pardoux have established the large population limit for a general epidemic model where the time spent by each individual in each compartment is drawn form an arbitrary distribution. This large population limit takes the form of a system of Volterra equations, also called distributed delay equations. These equations allow us to account for the “memory” of the epidemic dynamics, which for example affects the time it takes for public-health measures to have an impact on the growth of the number of infected individuals. We are also able to account for different behaviours (e.g. people who isolate themselves quickly versus people who don’t) whithout increas- ing the dimension of the system by choosing appropriate infectious period distributions (bimodal in this case). We then apply these results to the estimation of the state of the Covid-19 epidemic in France. Using the hospital data published by Sant´ePublique France, we are able to estimate the growth rate of the epidemic during its early stages (before lockdown measures, during national lockdown and after the easing of restric- tions). We then estimate the delays between infection and hospital deaths and deduce the corresponding estimate of the state of the Covid-19 epidemic throughout its early stages in different parts of the country. SLIGHT REDUCTION IN SARS-COV-2 EXPOSURE VIRAL LOAD DUE TO MASKING RESULTS IN A SIGNIFICANT REDUCTION IN TRANSMISSION WITH WIDESPREAD IMPLEMENTATION Ashish Goyal1, Daniel B. Reeves1, *E. Fabian Cardozo-Ojeda1, Bryan T. Mayer1, Joshua T. Schiffer1,2,3 1 Vaccine and Infectious Diseases Division, Fred Hutchinson Cancer Research Center 2 Department of Medicine, University of Washington, Seattle 3 Clinical Research Division, Fred Hutchinson Cancer Research Center *[email protected]

In the absence of vaccines, masks remain a vital tool for limiting SARS-CoV-2 spread in the population. However, the impact of masking on transmission within individual transmission pairs and at the population level is still under investigation. Here we utilize a mathematical model to quantitatively link mask efficacy to reductions in viral load and subsequent transmission risk. Our results reinforce that the use of masks by both a potential transmitter and exposed person substantially reduces the probability of successful transmission, even if masks only lower exposure viral load by ~50%. Under current average uptake estimates for the USA (75% of people wear masks 75% of the time), slight increases in masking relative to current levels would reduce the reproductive number substantially below 1, particularly if implemented comprehensively in potential super-spreader environments. Our model also predicts that moderately efficacious masks that reduce transmission risk by 50% will lower exposure viral load 10-fold among people who do get infected, potentially limiting infection severity. As an alternative intervention, we also assessed antiviral therapy. As peak viral load tends to occur before the onset of symptoms, antiviral therapy targeting symptomatic individuals had limited impact on transmission risk and population epidemics. Our model further projected that antiviral therapy is effective as a control measure only in a post-exposure prophylaxis setting, specifically if given to at least 50% of newly infected people within 3 days of an exposure. These results highlight the primacy of masking relative to other biomedical interventions under consideration for limiting the extent of the COVID-19 pandemic prior to widespread implementation of a vaccine.

A NETWORK SEIR MODEL: QUARANTINE EFFECTS AND GENETIC EVOLUTION Vitor M. Marquioni*1 and Marcus A. M. de Aguiar2 1 Ph.D. Student at Instituto de Física “Gleb Wataghin”, UNICAMP, Brazil. 2 Full Professor at Instituto de Física “Gleb Wataghin”, UNICAMP, Brazil.

The COVID-19 pandemics led the entire world to engage on a search for knowledge about this new disease in many different ways. Examples include building efficient predictive models, finding new drugs and vaccines and tracking the epidemic source with phylogenetic analyses. In this work, we developed an individual based SEIR model on networks that simulates the epidemic spreading and includes genetic information, so that the virus' evolution can be tracked. By changing the infection probability, we implemented a simple quarantine strategy, controlling its intensity, starting and duration time. Because the dynamic is stochastic, the same set of parameters leads to different outcomes every time the model is ran. We registered the number of times a particular set of parameters resulted in a good or a bad outcome. We showed that relatively short and intense quarantine periods can also be very effective in flattening the infection curve and even killing the virus, but the likelihood of such outcomes are low. Long quarantines of relatively low intensity, on the other hand, can delay the infection peak and reduce its size considerably with more than 50% probability, being a more effective policy than complete lockdown for short periods. We have also investigated the genetic evolution of the virus under a neutral evolutionary perspective. Our results point out that people traveling between weakly connected communities might be reinfected by more different (mutated) forms of the virus than in the case they are highly connected. We derived an equation describing the genetic evolution of the virus that can be added to the well-known SIR model. Real genome analysis also shows that this equation is in good agreement with the genetic evolution of SARS-CoV-2 epidemic in China.

*[email protected] PROCESS MEMORY IS KEY TO CAPTURING COVID-19 EPIDEMIOLOGICAL DYNAMICS Mircea T. Sofonea, Bastien Reyné, Baptiste Elie, Ramsès Djidjou-Demasse, Christian Selinger, Yannis Michalakis, *Samuel Alizon MIVEGEC, CNRS, IRD, Université de Montpellier,

SARS-CoV-2 virus has spread over the world creating one of the fastest pandemics ever. The absence of immunity, asymptomatic transmission, and the relatively high level of virulence of the COVID-19 infection it causes led to a massive flow of patients in intensive care units (ICU). This unprecedented situation calls for rapid and accurate mathematical models to best inform public health policies. Here, focusing on the epidemics in France, we show that accurately capturing the dynamics of incidence curves of ICU admissions and deaths requires to include memory effects in the model (also called ‘non-Markovian’). By developing an original statistical approach suited to our discrete-time approach, we estimate the value of the key epidemiological parameters, such as the basic reproduction number ($R_0$), and the efficiency of the national control strategy. These results have implications for the monitoring and control of COVID-19 epidemics worldwide.

Email : [email protected] CONTROLLING COVID-19 VIA TEST-TRACE-QUARANTINE

1 1† 2,3† 1 1 *Cliff C. Kerr ,​ Dina Mistry ,​ Robyn M. Stuart ,​ Katherine Rosenfeld ,​ Gregory R. Hart ,​ Rafael 1 ​ ​ 1 ​ 1 ​ 3 ​ 1 C. Núñez ,​ Prashanth Selvaraj ,​ Jamie A. Cohen ,​ Romesh G. Abeysuriya ,​ Lauren George ,​ ​ 1 ​ 4 ​ 5 ​ 5 ​ Brittany Hagedorn ,​ Michał Jastrzębski ,​ Meaghan Fagalde ,​ Jeffrey Duchin ,​ Michael 1 ​ 1 ​ ​ ​ Famulare ,​ and Daniel J. Klein ​ ​

1 ​ Institute for Disease Modeling, Seattle, WA, USA 2 ​ Department of Mathematical Sciences, University of Copenhagen, Copenhagen, Denmark 3 ​ Burnet Institute, Melbourne, VIC, Australia 4 ​ GitHub, Inc., San Francisco, CA, USA 5 ​ Seattle-King County Health Authority, Seattle, WA, USA

* Presenting author. E-mail: [email protected] ​ † Contributed equally

Initial COVID-19 containment efforts in the United States largely focused on physical distancing, including school and workplace closures. However, these interventions have come at an enormous societal and economic cost. Here we explore the feasibility of alternative control strategies, focusing on "test-trace-quarantine" (TTQ): routine testing of primarily symptomatic individuals, tracing and testing their known contacts, and placing their contacts in quarantine.

We performed this analysis using Covasim, an open-source agent-based model, calibrated to detailed demographic, mobility, and epidemiological data for the Seattle region. Covasim is designed to capture the nuances of realistic COVID-19 transmission, including: age and population structure; transmission networks in different social layers, including households, schools, workplaces, communities, and aged care facilities; age-specific susceptibility, symptomaticity, and mortality; and simple intrahost viral dynamics. Covasim also supports an extensive set of interventions, namely: non-pharmaceutical interventions, such as distancing and masks; testing interventions, such as symptomatic and asymptomatic testing, contact tracing, isolation, and quarantine; and pharmaceutical interventions, such as therapeutics and vaccines. These interventions can incorporate the effects of delays, micro-targeting, and other factors.

We found that for the Seattle setting, with high mask compliance and schools remaining closed, realistic levels of testing and tracing were sufficient to maintain epidemic control under a return to full workplace and community mobility. We validated these findings against preliminary programmatic and epidemiological data from Seattle after the TTQ program was implemented, and compare to international settings with epidemic control, including Australia and South Korea. While TTQ is shown to be capable of controlling the epidemic in both theory and practice, its success is contingent on relatively short testing and tracing delays, high compliance with quarantine, and moderate to high mask use. Thus, strong performance in each aspect of the program must be maintained in order to control transmission. ESTIMATION OF SARS-COV-2 MORTALITY DURING THE EARLY STAGES OF AN EPIDEMIC: A MODELING STUDY IN HUBEI, CHINA, AND SIX REGIONS IN EUROPE

Anthony Hauser1,*, Michel J. Counotte1, Charles C. Margossian2, Garyfallos Konstantinoudis3, Nicola Low1, Christian L. Althaus1, Julien Riou1,4

1 Institute of Social and Preventive Medicine, University of Bern, Bern, Switzerland 2 Department of Statistics, Columbia University, New York, New York, United States of America 3 MRC Centre for Environment and Health, Department of Epidemiology and Biostatistics, School of Public Health, Imperial College London, London, United Kingdom 4 Division of infectious diseases, Federal Office of Public Health, Bern, Switzerland * presenting author, email address: [email protected]

Abstract:

Reliable estimates of mortality from SARS-CoV-2 infection are essential for understanding clinical prognosis, planning healthcare capacity, and epidemic forecasting. The case–fatality ratio (CFR), calculated from total numbers of reported cases and reported deaths, is the most commonly reported metric, but it can be a misleading measure of overall mortality.

We developed an age-stratified susceptible-exposed-infected-removed (SEIR) compartmental model to (1) simulate the transmission dynamics of SARS-CoV-2 using publicly available surveillance data and (2) infer estimates of SARS-CoV-2 mortality adjusted for biases. Our model accounts for two biases: preferential ascertainment of severe cases and right-censoring of mortality. We fitted the transmission model to surveillance data from Hubei Province, China, and applied the same model to six regions in Europe: Austria, Bavaria (Germany), Baden-Württemberg (Germany), Lombardy (Italy), Spain, and Switzerland. In Hubei, the baseline estimates were as follows: CFR 2.4% (95% credible interval [CrI] 2.1%– 2.8%), and infection-fatality ratio (IFR) 2.9% (2.4%–3.5%). Across the six locations in Europe, estimates of CFR varied widely. Estimates and IFR, adjusted for bias, were more similar to each other but still showed some degree of heterogeneity. Estimates of IFR ranged from 0.5% (95% CrI 0.4%–0.6%) in Switzerland to 1.4% (1.1%–1.6%) in Lombardy, Italy. In all locations, mortality increased with age. Among individuals 80 years or older, estimates of the IFR suggest that the proportion of all those infected with SARS-CoV-2 who will die ranges from 20% (95% CrI 16%–26%) in Switzerland to 34% (95% CrI 28%–40%) in Spain.

The IFR, adjusted for right-censoring and preferential ascertainment of severe cases, provides more reliable estimates of SARS-CoV-2 mortality than the CFR. Geographic differences in IFR suggest that a single IFR should not be applied to all settings to estimate the total size of the SARS-CoV-2 epidemic in different countries.

HOSPITAL DEMAND UNDER INTERVENTION DURING THE COVID-19 PANDEMIC : A MODELING STUDY

Katsuma Hayashi*1, Taishi Kayano1 , Sumire Sorano 2 ,Hiroshi Nishiura1 School of 1Public Health and Graduate School of Medicine, Kyoto University hayashi.katsuma.7w@kyoto- u.au.jp (K.H.); [email protected] (T.K.); [email protected] (H.N.) 2London School of Hygiene and Tropical Medicine, Keppel Street, London, United Kingdom; [email protected] (S.S.)

Background: In late March 2020, there was a spike in hospitalizations in Japan. The incidence of coronavirus disease (COVID-19) decreased temporarily from March to May as a result of host and hostess club closures, reduced restaurant hours, and demands for reduced spontaneous contact outside the home. Managing the hospital demand was important to prepare for the second wave, and the scenario analysis required for planned hospital bed interventions.

Methods: we analyzed the data of the first wave by age group and made it possible to predict the number of hospitalizations by the age of the second wave. We analyzed two different areas because the age patterns of epidemics differ in urban areas and elsewhere. Osaka, which accounts for the majority of young people, and Hokkaido, which accounts for the majority of inpatients.

Results:By estimating the exponential growth rates of cases by age group and assuming the expected decrease in those rates under intervention, we obtained the predicted epidemic curve of cases in addition to hospitalization. We have shown that the later the intervention, the higher the peak of hospitalization.

Conclusion:Although the approach relies on a simplistic model, the proposed framework can guide local government to secure the essential number of hospital beds for COVID-19 cases and formulate action plans.

TIME TREND OF CASE FATALITY RISK OF COVID-19 IN JAPAN

*Taishi Kayano1, Sung-Mok Jung1,2, Hiroshi Nishiura1 1 School of Public Health and Graduate School of Medicine, Kyoto University 2 Graduate Scholl of Medicine, Hokkaido University

Background: Although the case fatality risk (CFR) of COVID-19 have been discussed all over the world, the estimate of that needs to be paid attention due to the nature of the data of the ongoing epidemic. The time trend of CFR can describe not only the situation but also the impact of interventions throughout the epidemic. The present study estimated the adjusted confirmed CFR of COVID-19 by age groups in Japan, analysing the change in those values over the course of the epidemic.

Methods: We investigated the confirmed COVID-19 cases and deaths in Japan from January to July 2020. To avoid biased CFR, the numerator (i.e. deaths) was adjusted by the probability mass function of the delay from illness onset to death. Maximum likelihood estimation was performed to estimate the unbiased CFR according to age groups by every month.

Results: Estimated CFR in all age groups were significantly increased in the early stage of the first wave in Japan. However, as the peak passed, CFR started decreasing in all age groups. Those who were aged 60 and over had more compelling changes compared to younger age groups. The estimated CFR in July when was the part of the early stage of the second wave were much smaller than those in April when we saw the similar trend of the escalated incidences.

Conclusion: CFR in all age groups did not consist along in months in Japan. After abrupt increased CFR in March and April, they significantly declined in the following months, implying the improvement of interventions in medical facilities and/or active surveillance.

Email: [email protected]

PHENOMENOLOGICAL AND MECHANISTIC MODEL COMPARISON FOR THE FORECAST OF AN EPIDEMIC USING COVID-19 REPORTED DATA IN CHINA *Takeshi Miyama1,2, Katsuma Hayashi2, Sung-mok Jung2, Asami Anzai2, Ryo Kinoshita2, Tetsuro Kobayashi2, Natalie M. Linton2, Ayako Suzuki2, Taishi Kayano2, Andrei R. Akhmetzhanov3 and Hiroshi Nishiura2 1Osaka Institute of Public Health, Osaka, 2Kyoto University, 3National Taiwan University

Forecasting the future incidence of an epidemic helps make decisions for interventions. During the early epidemic period, however, limited information is available, and building detailed mechanistic models reflecting population behaviour is somewhat challenging. While phenomenological models, which can be applied with the daily number of new cases/cumulative cases only, could capture the trend of the epidemic. In this study, the Richards model and the approximate solution of the basic SIR differential equations were employed as phenomenological models using the COVID-19 epidemic data in China. Also, simple mechanistic models with the intensive nationwide intervention, the lockdown, were built. Specifically, the exponential growth model and the SIR model were built incorporating the lockdown effect by changing the growth rate/transmission parameter before and after the lockdown. Root mean square errors during both calibration and forecasting periods were calculated for each model to compare the validity of the models above. Also, the calibration periods were changed to assess forecasting capacity. The results will be reported at the presentation, and the characteristics of the models in the case of COVID-19 in China will be discussed.

*Takeshi Miyama, Osaka Institute of Public Health, Nakamichi 1-3-69, Higashinari, Osaka 537- 0025, Japan. Email: [email protected] GLOBAL GEOGRAPHIC AND TEMPORAL ANALYSIS OF SARS-COV-2 HAPLOTYPES NORMALIZED BY COVID-19 CASES

*Santiago Justo Arévalo1,2,, Daniela Zapata Sifuentes1,±, César Huallpa Robles3,±, Gianfranco Landa Bianchi1,±, Adriana Castillo Chávez1,±, Romina Garavito-Salini Casas1,±, Guillermo Uceda-Campos2,4 Roberto Pineda Chavarría1.

1.- Universidad Ricardo Palma, Facultad de Ciencias Biológicas, Lima – Perú. 2.- Universidade de Sao Paulo, Instituto de Química, Departamento de Bioquímica, São Paulo - Brasil 3.- Universidad Nacional Agraria La Molina, Facultad de Ciencias, Lima – Perú. 4.- Universidad Nacional Pedro Ruiz Gallo, Facultad de Ciencias Biológicas, Lambayeque - Perú ± These authors contributed equally to this work. * Corresponding author: [email protected]

ABSTRACT:

COVID-19 was declared a pandemic by the World Health Organization on March. Since January a high number of genomes began to be sequenced around the world. This marks a unique opportunity to analyze virus spreading and evolution in a worldwide context. Identification of mutations rising in all around the world is important track virus adaptation and evolution. However, differences in the number of sequenced genomes between countries and/or months make it difficult to identify the emergence of mutations in regions where few genomes are sequenced but a large number of cases are reported. We proposed an approach based on the normalization by COVID-19 cases of relative frequencies of mutations using all the available data to identify most frequent mutations. Thus, we can use a similar normalization approach to tracking the global temporal and geographic distribution in the world. Using 48 776 genomes, we identify 5 major haplotypes based on 9 high-frequency mutations. Normalized global geographic and temporal analysis is presented here highlighting the current importance of nucleocapsid mutations (R203K, G204R) above the highly discussed D614G in spike protein. Also, we analyzed age, gender, and patient status distribution by haplotypes, but scarce and not well- organized information about this is publicly available. For that, we create a web-service to continuously update our normalized analysis of mutations and haplotypes, and to allow researchers to voluntarily share patient status information in a well-organized manner to improve analyses and making possible monitor the emergence of mutations and/or haplotypes with patients preferences or different pathogenic features. Finally, we discuss currently structural and functional hypotheses in the most frequently identified mutations.

The unintended consequences of inconsistent pandemic control policies Benjamin M. Althouse1,2,3,*, Brendan Wallace4, Brendan Case5,6, Samuel V. Scarpino7,8,9, Andrew M. Berdahl10, Easton R. White11,12, and Laurent Hebert-Dufresne´ 5,6

1Institute for Disease Modeling, Bellevue, WA 98005 2University of Washington, Seattle, WA 98105 3New Mexico State University, Las Cruces, NM 88003 4Department of Applied Mathematics, University of Washington, Seattle, WA 98195, USA 5Department of Computer Science, University of Vermont, Burlington, VT 05405, USA 6Vermont Complex Systems Center, University of Vermont, Burlington, VT 05405, USA 7Network Science Institute, Northeastern University, Boston, MA, USA 8ISI Foundation, Turin, Italy 9Santa Fe Institute, Santa Fe, NM, USA 10School of Aquatic & Fishery Sciences, University of Washington, Seattle, WA 98195, USA 11Department of Biology, University of Vermont, Burlington, VT 05405, USA 12Gund Institute for Environment, University of Vermont, Burlington, VT 05405, USA *[email protected]

ABSTRACT

Controlling the spread of COVID-19 – even after a licensed vaccine is available – requires the effective use of non- pharmaceutical interventions, e.g., physical distancing, limits on group sizes, mask wearing, etc.. To date, such interventions have neither been uniformly nor systematically implemented in most countries. For example, even when under strict stay- at-home orders, numerous jurisdictions granted exceptions and/or were in close proximity to locations with entirely different regulations in place. Here, we investigate the impact of such geographic inconsistencies in epidemic control policies by coupling search and mobility data to a simple mathematical model of SARS-COV2 transmission. Our results show that while stay-at-home orders decrease contacts in most areas of the United States of America (US), some specific activities and venues often see an increase in attendance. Indeed, over the month of March 2020, between 10 and 30% of churches in the US saw increases in attendance; even as the total number of visits to churches declined nationally. This heterogeneity, where certain venues see substantial increases in attendance while others close, suggests that closure can cause individuals to find an open venue, even if that requires longer-distance travel. And, indeed, the average distance travelled to churches in the US rose by 13% over the same period. Strikingly, our mathematical model reveals that, across a broad range of model parameters, partial measures can often be worse than no measures at all. In the most severe cases, individuals not complying with policies by traveling to neighboring jurisdictions can create epidemics when the outbreak would otherwise have been controlled. Taken together, our data analysis and modelling results highlight the potential unintended consequences of inconsistent epidemic control policies and stress the importance of balancing the societal needs of a population with the risk of an outbreak growing into a large epidemic.

Keywords: COVID-19, SARS-COV2, social distancing, non-pharmaceutical interventions, human behavior

Corresponding author: Benjamin M Althouse Institute for Disease Modeling 3150 139th Ave SE Bellevue, WA, 98005 Phone: (425) 777-9615 Email: [email protected] medRxiv preprint doi: https://doi.org/10.1101/2020.05.04.20090092.this version posted August 1, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY-NC 4.0 International license .

Variation in SARS-CoV-2 free-living survival and environmental transmission can modulate the intensity of emerging outbreaks

C. Brandon Ogbunugafor1,2,3*, Miles D. Miller-Dickson2, Victor A. Meszaros2, Lourdes M. Gomez1,2, Anarina L. Murillo4,5, and Samuel V. Scarpino6

1Department of Ecology and Evolutionary Biology, Yale University 06520 2Department of Ecology and Evolutionary Biology, Brown University 02912 3Center for Computational Molecular Biology, Brown University 02912 4Department of Pediatrics, Warren Alpert Medical School at Brown University 02912 5Center for Statistical Sciences, Brown University School of Public Health 02903 6Network Science Institute, Northeastern University 02115

Keywords: Environmental transmission, indirect transmission, fomites, viral free-living survival, emerging infectious diseases, ecology of infectious diseases, coronaviruses, mathematical modeling

*Correspondence to: C. Brandon Ogbunugafor Department of Ecology and Evolutionary Biology Yale University [email protected]

1 medRxiv preprint doi: https://doi.org/10.1101/2020.05.04.20090092.this version posted August 1, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY-NC 4.0 International license .

Abstract Variation in free-living, microparasite survival can have a meaningful impact on the

ecological dynamics of established and emerging infectious diseases. Nevertheless,

resolving the importance of environmental transmission in the ecology of epidemics

remains a persistent challenge, requires accurate measuring the free-living survival of

pathogens across reservoirs of various kinds, and quantifying the extent to which

interaction between hosts and reservoirs generates new infections. These questions are

especially salient for emerging pathogens, where sparse and noisy data can obfuscate

the relative contribution of different infection routes. In this study, we develop a

mechanistic, mathematical model that permits both direct (host-to-host) and indirect

(environmental) transmission and then fit this model to empirical data from 17 countries

affected by an emerging virus (SARS-CoV-2). From an ecological perspective, our

model highlights the potential for environmental transmission to drive complex, non-

linear dynamics during infectious disease outbreaks. Summarizing, we propose that

fitting such models with environmental transmission to real outbreak data from SARS-

CoV-2 transmission highlights that variation in environmental transmission is an

underappreciated aspect of the ecology of infectious disease, and an incomplete

understanding of its role has consequences for public health interventions.

2

ASSESSING LINKED SELECTION AND LONG-DISTANCE ASSOCIATIONS OF FUNCTIONAL MUTATIONS IN SARS-COV2 VARIANTS IN INDIA 1 2 *​ Ishaan Singh Rawal, ​ A​ run Sethuraman 1 ​ Department of Biological Sciences, Birla Institute of Technology and Science, Pilani 333031, India (E-mail: [email protected]) 2 ​ Department of Biological Sciences, California State University San Marcos, San Marcos, CA 92096, USA

The ongoing COVID-19 pandemic is caused by a novel strain of coronavirus, SARS-CoV-2. As reported, the RNA virus is mutating. The mutations and their effects need to be taken into account when designing the vaccines or other therapeutics. Motivated by this, we analyzed 1,202 full-genome sequences of the virus sequenced in India. We filtered non-synonymous mutations on two levels- first on the basis of the percentage identity, followed by the divergence in the biochemical nature of the mutated amino acids, and identified 15 functional mutations, including the previously characterized D616G that controls structural stability of the spike protein. However, some mutations, like the A994D in the nsp3 protein, cannot be explained due to the lack of knowledge about the protein. These characterized mutations, which are distributed throughout the genome, show a distinct pattern of clustering into two distinct groups- (nsp1-nsp5) group and (n-s) group, separated by a highly conserved region of 10,500 base pairs.

We hypothesize that selective pressure on the virus can lead to the association of both- intra-cluster and inter-cluster mutations. This is different from the conventional selective sweeps, which is based on the hitchhiking of closely linked genes (in recombining species). Hence, this heterogeneous pattern of mutations cannot be explained by the standard LD and SFS approaches. Here we design and apply a series of long-distance association tests to quantify linked selection along the distant functional mutations on the SARS-CoV-2 genome. These tests could, in principle, provide clues about functional mutations on unknown segments of the genome, the mechanisms of mutation and the evolutionary trajectory of the virus. 1 Crowding and the shape of COVID-19 epidemics 2 3 Benjamin Rader1,2,*, Samuel V. Scarpino3,4,5,*,$, Anjalika Nande6, Alison L. Hill6, Ben Adlam6, Robert C. 4 Reiner7,8, David M. Pigott7,8, Bernardo Gutierrez9,10, Alexander Zarebski9, Munik Shrestha3, Open 5 COVID-19 Data Working Group#, John S. Brownstein1,11, Marcia C. Castro12, Christopher Dye9, Huaiyu 6 Tian13, Oliver G. Pybus9,14,$, Moritz U.G. Kraemer9,$ 7 8 1. Computational Epidemiology Lab, Boston Children’s Hospital, Boston, United States 9 2. Department of Epidemiology, Boston University School of Public Health, Boston, United States 10 3. Network Science Institute, Northeastern University, Boston, United States 11 4. ISI Foundation, Turin, Italy 12 5. Santa Fe Institute, Santa Fe, United States 13 6. Program for Evolutionary Dynamics, Harvard University, Cambridge, United States 14 7. Department of Health Metrics, University of Washington, Seattle, United States 15 8. Institute for Health Metrics and Evaluation, University of Washington, Seattle, United States 16 9. Department of Zoology, University of Oxford, Oxford, United Kingdom 17 10. School of Biological and Environmental Sciences, Universidad San Francisco de Quito USFQ, 18 Quito, Ecuador 19 11. Harvard Medical School, Boston, United States 20 12. Department of Global Health and Population, Harvard T.H. Chan School of Public Health, 21 Boston, United States 22 13. State Key Laboratory of Remote Sensing Science, College of Global Change and Earth System 23 Science, Beijing Normal University, Beijing, China 24 14. Department of Pathobiology and Population Science, The Royal Veterinary College, London, 25 United Kingdom 26 27 *contributed equally as first authors 28 $correspondence should be addressed to [email protected], [email protected] and 29 [email protected] 30 #Members of the Open COVID-19 Data Working Group are listed at the end of the manuscript 31

1 32 Abstract 33 The COVID-19 pandemic is straining public health systems worldwide and major non- 34 pharmaceutical interventions have been implemented to slow its spread1–4. During the initial phase 35 of the outbreak, dissemination of SARS-CoV-2 was primarily determined by human mobility from 36 Wuhan5,6. Yet empirical evidence on the effect of key geographic factors on local epidemic 37 transmission is lacking7. We analyse highly-resolved spatial variables in cities together with case 38 count data in order to investigate the role of climate, urbanization, and variation in interventions. 39 We show that the degree to which cases of COVID-19 are compressed into a short period of time 40 (peakedness of the epidemic) is strongly shaped by population aggregation and heterogeneity, such 41 that epidemics in crowded cities are more spread over time, and crowded cities have larger total 42 attack rates than less populated cities. Observed differences in the peakedness of epidemics are 43 consistent with a metapopulation model of COVID-19 that explicitly accounts for spatial 44 hierarchies. We pair our estimates with globally-comprehensive data on human mobility and 45 predict that crowded cities worldwide could experience more prolonged epidemics. 46 47 Main 48 Predicting the epidemiology of the COVID-19 pandemic is a priority for guiding epidemic responses 49 around the world. China has undergone its first epidemic wave and, remarkably, cities across the country 50 are now reporting few or no locally-acquired cases8. Analyses have indicated that that the spread of 51 COVID-19 from Hubei to the rest of China was driven primarily by human mobility from Wuhan6,9, and 52 that the stringent measures to restrict human movement and public gatherings within and among cities in 53 China were associated with bringing local epidemics under control5. Key uncertainties remain as to which 54 geographic factors drive the local transmission dynamics of COVID-19 and initial analysis suggests a 55 limited role of climate in determining epidemic growth10. 56 57 Spatial heterogeneity in infectious disease transmission can be influenced by local differences in 58 population or human movements, such that high local population densities might catalyse the spread of 59 novel pathogens due to higher contact rates with susceptible individuals11,12. For respiratory pathogens, 60 the temporal clustering of cases in an epidemic (i.e., the shortest period during which the majority of 61 cases are observed) varies with increased indoor crowding and socio-economic and climatic factors13–18. 62 The temporal concentration of cases is minimized when incidence is spread evenly across time and 63 increases as incidence becomes more concentrated in particular days, as has been observed for 64 influenza13. In any given location, a higher temporal concentration of cases may require a larger surge

2 APPLYING V-PIPE TO SARS-COV-2 *Ivan Topolsky 1,2 Aashil Batavia 1,2,3 Fritz Bayer 1,2 Nico Borgsmüller 1,2 Arthur Dondi 1,2 Monica Drăgan 1,2 Pedro Ferreira 1,2 Kim Philipp Jablonski 1,2 Katharina Jahn 1,2 Jack Kuipers 1,2 Lisa Lambretti 1,2 Martin Pirkl 1,2 Susana Posada-Céspedes 1,2 Prof. Niko Beerenwinkel 1,2

1. Computational Biology Group - ETH Zürich 2. Swiss Institute of Bioinformatics 3. UniSpitalZürich

Next-generation sequencing (NGS) of viral genomes is now the method of choice for analysing the diversity of intra- and inter- host virus populations, including epidemiological studies and individual treatment optimization in clinical virology.

V-pipe provides a data analysis workflow for clinical applications of NGS data obtained from viral genomes. The pipeline assesses data quality, performs read alignment and infers intra-sample viral genomic diversity: single-nucleotide variations and haplotype assembly at both a local and global level.

In this presentation we show the latest development of our SIB (Swiss Institute of Bioinformatics) resource: V-Pipe.

V-pipe is used to fully automate the processing of raw reads data produced by sequencing effort on samples obtained from SARS-CoV-2 positive viral swab-tests done by Viollier AG in Basel. Sequencing has been performed at the Genomics Facility Basel, on an Illumina MiSeq machine, and our team was provided with the raw reads over an OpenBIS database.

In addition to that, our team has also focused on the visualization and reporting generation to assist the analysis of intra-host and whole cohort diversity of virus variations.

V-pipe has performed quality checks automatically at each step of the pipeline. Passing sequences have then been uploaded to repositories such as GISAID and ENA (project PRJEB38472), enabling downstream phylogenetic analysis by other members of the task force such as the Swiss build of Nextstrain. These results are available at: https://ncs-tf.ch/en/nextstrain

V-pipe has also produced reports that enable easy exploration of SNV and visually situating them in regions of interest. The reports are produced in an HTML and JavaScript format, making it easy for the researcher to display them into a standard web browser. E-mails: Presenter: [email protected] Corresponding author: [email protected]

External links: Dedicate SARS-CoV-2 sub-page of V-pipe's website: https://cbg-ethz.github.io/V-pipe/sars-cov-2/

Informations about V-pipe are available at: https://cbg-ethz.github.io/V-pipe/

Illustrations: Genomics Facility Basel Downstream Phylogenetic RECONSTRUCTING THE EARLY GLOBAL DYNAMICS OF UNDER-ASCERTAINED COVID-19 CASES AND INFECTIONS

Authors: Timothy W. Russell1*, Nick Golding2, Joel Hellewell1, Sam Abbott1, W. John Edmunds1, Adam J. Kucharski1 on behalf of the CMMID COVID-19 working group

1Centre for Mathematical Modelling of Infectious Diseases, London School of Hygiene & Tropical Medicine 2Telethon Kids Institute and Curtin University, Perth, Western Australia

Asymptomatic or subclinical SARS-CoV-2 infections are often unreported, which means that confirmed case counts may not accurately reflect underlying epidemic dynamics. Understanding the level of ascertainment (the ratio of confirmed symptomatic cases to the true number of symptomatic individuals) and undetected epidemic progression is crucial to informing COVID-19 response planning, including the introduction and relaxation of control measures. Estimating case ascertainment over time allows for accurate estimates of specific outcomes such as seroprevalence, which is essential for planning control measures. Using reported data on COVID-19 cases and fatalities globally, we estimated the proportion of symptomatic cases (i.e. any person with any of fever >= to 37.5C, cough, shortness of breath, sudden onset of anosmia, ageusia or dysgeusia illness) that were reported in 210 countries and territories, given those countries had experienced more than ten deaths. We used published estimates of the case fatality ratio (CFR) as an assumed baseline. We then calculated the ratio of this baseline CFR to an estimated local delay-adjusted CFR to estimate the level of under-ascertainment in a particular location. We then fit a Bayesian Gaussian process model to estimate the temporal pattern of under-ascertainment. We estimate that, during March 2020, the median percentage of symptomatic cases detected across the 84 countries which experienced more than ten deaths ranged from 2.38% (Bangladesh) to 99.6% (Chile). Across the ten countries with the highest number of total confirmed cases as of 6th July 2020, we estimated that the peak number of symptomatic cases ranged from 1.4 times (Chile) to 17.8 times (France) larger than reported. Comparing our model with national and regional seroprevalence data where available, we find that our estimates are consistent with observed values. Finally, we estimated seroprevalence for each country. Despite low case detection in some countries, our results that adjust for this still suggest that all countries have had only a small fraction of their populations infected as of July 2020. We found substantial under-ascertainment of symptomatic cases, particularly at the peak of the first wave of the SARS-CoV-2 pandemic, in many countries. Reported case counts will therefore likely underestimate the rate of outbreak growth initially and underestimate the decline in the later stages of an epidemic. Although there was considerable under-reporting in many locations, our estimates were consistent with emerging serological data, suggesting that the proportion of each country's population infected with SARS- CoV-2 worldwide is generally low COUNTERINTUITIVE COVID-19 MORTALITY-MOBILITY CORRELATIONS ACCOUNTED FOR BY UNDERLYING PRE-LOCKDOWN TRANSMISSION RATES

Daniel D. Lewis, Michael Pablo, Xinyue Chen, Rob Rodick, *Leor Weinberger [email protected], [email protected], [email protected], [email protected], [email protected]

Gladstone Institute of Virology Gladstone|UCSF Center for Cell Circuitry Department of Biochemistry and Biophysics Department of Pharmaceutical Chemistry University of California, San Francisco

Recent reports argue that COVID-19 lockdowns ‘failed’, with mortality perversely increasing with lockdown intensity (1). We analyze mortality dynamics at 1,114 locations at three epidemiological scales (102 countries, 50 states, and 962 counties) against Google mobility data for each locale (at workplace, transit, commercial, parks, and residential areas). We find that mobility and death have a complex association, and that careful consideration is needed to assess the impact of COVID-19 lockdowns. Consistent with classical theory (Kermack-McKendrick, 1927), the data show that COVID-19 mortality strongly correlates with a region’s total population and obscures the impact of mobility changes. Early mobility changes weakly correlate with reduced per-capita mortality but certain high-mortality regions exhibit reductions in mortality rate before the effects of decreased mobility could manifest (a.k.a., early-phase decoupling). While overall mobility reductions did correlate with increased mortality, epidemiological model regression (>7x107 simulations) identified a class of models that fit the early-phase decoupling data and shows that the ‘perverse’ mobility-mortality correlation is explained by underlying transmission rates: specifically, locales with faster pre-lockdown transmission exhibited greater mobility reductions. Overall, the analysis identifies models appropriate for analyzing the impact of COVID-19 interventions and, critically, explains the seemingly counterintuitive mobility-mortality correlation without a causal relationship or perverse mechanism.

(1) Luskin, D. “The Failed Experiment of Covid Lockdowns” Wall Street Journal. Sept. 1st, 2020 32. GENOMIC EPIDEMIOLOGY OF THE EARLY STAGES OF SARS-COV-2 OUTBREAK IN RUSSIA, *G. Bazykin DIFFERING IMPACTS OF GLOBAL AND REGIONAL RESPONSES ON SARS-COV-2 TRANSMISSION CLUSTER DYNAMICS

*Brittany Rife Magalis,1;2_ Andrea Ramirez-Mata,1;2 Anna Zhukova,3 Carla Mavian,1;2 Simone Marini,2;4 Frederic Lemoine,3 Mattia Prosperi,4 Olivier Gascuel,3 Marco Salemi1;2 1Department of Pathology, Immunology, and Laboratory Medicine, University of Florida, Gainesville, Florida, 32610, USA 2Emerging Pathogens Institute, University of Florida, Gainesville, Florida, 32610, USA 3Department of Computational Biology, Institut Pasteur, Paris, 75015, France 4Department of Epidemiology, University of Florida, Gainesville, Florida, 32610, USA

Although the global response to COVID-19 has not been entirely unified, the opportunity arises to assess the impact of regional public health interventions and to classify strategies according to their outcome. Analysis of genetic sequence data gathered over the course of the pandemic allows us to link the dynamics associated with networks of connected individuals with specific interventions. In this study, clusters of transmission were inferred from a phylogenetic tree representing the relationships of patient sequences sampled from December 30, 2019 to April 17, 2020. Metadata comprising sampling time and location were used to define the global behavior of transmission over this earlier sampling period, but also the involvement of individual regions in transmission cluster dynamics. Results demonstrate a positive impact of international travel restrictions and nationwide lockdowns on global cluster dynamics. However, residual, localized clusters displayed a wide range of estimated initial secondary infection rates, for which uniform public health interventions are unlikely to have sustainable effects. Our findings highlight the presence of so-called "super-spreaders", with the propensity to infect a larger-than- average number of people, in countries, such as the USA, for which additional mitigation efforts targeting events surrounding this type of spread are urgently needed to curb further dissemination of SARS-CoV-2.

E-mail: [email protected].

FEASIBILITY OF CONTAINING COVID-19 BY SYMPTOM BASED CASE ISOLATION

*Ryo Kinoshita1, Asami Anzai1, Sung-mok Jung1, Natalie M Linton1, Takeshi Miyama2, Tetsuro Kobayashi1, Katsuma Hayashi1, Ayako Suzuki1, Yichi Yang3, Andrei R Akhmetzhanov3, and Hiroshi Nishiura1 1Kyoto University, School of Public Health 2Osaka Institute of Public Health 3Graduate School of Medicine, Hokkaido University

When a novel infectious disease emerges, enhanced contact tracing and isolation is a key to prevent a major epidemic. This strategy has been successful for the control of severe acute respiratory syndrome (SARS) and Middle East respiratory syndrome (MERS) without causing a global pandemic. Considering that asymptomatic and pre-symptomatic transmission are substantial for novel coronavirus disease (COVID-19), the feasibility of preventing the major epidemic has been questioned.

Using a two-type branching process model of a negative binomial distribution that account for symptomatic and asymptomatic infection, the present study assesses the feasibility of containing COVID-19 by computing the probability of a major epidemic given untraced symptomatic cases. We assumed that contact tracing reduces secondary transmission only among symptomatic cases after the symptom onset. A broad range of asymptomatic ratio and relative infectiousness among asymptomatic cases were assumed.

If the number of asymptomatic transmissions are substantial, cutting chains of transmission by means of contact tracing and case isolation of symptomatic individuals would be very challenging. While contact tracing and isolation against symptomatic cases contribute to advance the feasibility of containment, even if isolation of symptomatic cases is conducted swiftly after the symptom onset, only secondary transmissions after the symptom onset can be prevented. Our results highlights the importance of active contact tracing among close contacts of an infected individual to identify and isolate asymptomatic and pre-symptomatic cases.

Email: [email protected] (word count 278/300) OPTIMISING TIME-LIMITED NON-PHARMACEUTICAL INTERVENTIONS FOR COVID-19 OUTBREAK CONTROL Alex L.K Morgan*1, Mark E.J Woolhouse2, Graham F. Medley3 and Bram A.D van Bunnik2

1Centre for Immunity, Infection & Evolution and School of Biological Sciences, University of Edinburgh, Edinburgh, United Kingdom 2Usher Institute, University of Edinburgh, Edinburgh, United Kingdom 3Centre for Mathematical Modelling of Infectious Diseases and Department of Global Health & Development, London School of Hygiene and Tropical Medicine, London, United Kingdom

Correspondance Email: [email protected]

Retrospective analyses into the non-pharmaceutic interventions (NPIs) used to combat the ongoing COVID-19 outbreak has highlighted the potential of optimising interventions. These optimal interventions allow for policy makers to manage NPIs in order to minimise the epidemiological and human health impacts of both COVID-19 and the intervention itself. Here, we use a susceptible- infectious-recovered (SIR) mathematical model to explore the feasibility of optimising the duration, magnitude and trigger point of five different NPI scenarios to minimise the peak prevalence and attack rate of a simulated UK COVID-19 outbreak.

An optimal parameter space to minimise the peak prevalence and the attack rate was identified for each intervention scenario, with each scenario differing with regards to how reductions to transmission were modelled. However, we show that these optimal interventions are fragile, sensitive to epidemiological uncertainty and prone to implementation error. We highlight the use of suboptimal interventions as a more robust alternative. While not as effective as optimal interventions, suboptimal interventions are still capable of reducing peak prevalence and attack rate over a wider, more achievable parameter space. This work provides an illustrative example of the concept of intervention optimisation across a wide range of different NPI strategies.