correspondence

The lag in SARS-CoV-2 genome submissions to GISAID

To the Editor — Genomic surveillance krisp.org.za/ngs-sa/ngs-sa_network_for_ genomes each, which is 1.5% and 9.3% of of the evolving SARS-CoV-2 strains is genomic_surveillance_south_africa/) and their respective positive samples. India, an important tool for helping control the SPHERES (United States; https://www.cdc. being the second-largest country on the pandemic1. For efficient surveillance, the gov/coronavirus/2019-ncov/covid-data/ basis of both total population and known first major requirement for analysis of how spheres.html). Even so, although an COVID-19 cases, has sequenced a mere the virus is evolving and spreading is the increasing fraction of COVID-19 samples 0.05% of the collected samples. On average, availability of all sequenced genomes on are being sequenced9, an equally important African, Asian and South American an open-access platform that is accessible issue is how soon the sequences are being countries have sequenced a mere 0.36%, to researchers worldwide. Therefore, submitted to GISAID. Rapid submission 0.21% and 0.07% of their total COVID-19 soon after researchers became aware of is important as it enables the international samples, respectively, whereas this number COVID-19, toward the end of 2019, the community to analyze the variants is 1.9%, 1.4% and 37% for European, Global Initiative on Sharing All emerging around the world quickly North American and Oceania countries. Data (GISAID), an existing platform for and provide actionable information to Population-wise, most of the European sharing influenza virus sequences, began governments. countries, the United States, Israel (Asia) receiving deposits of SARS-CoV-2 genome Our statistical analysis (Fig. 1, and the island of Réunion have sequenced sequences. Here we report an analysis of Supplementary Fig. 1 and Supplementary samples from over 1,000 people per the median collection to submission time Tables 1 and 2) for the 1,718,035 million population (ppmp). Among (CST) lag for SARS-CoV-2 sequences to SARS-CoV-2 strains submitted to GISAID countries with over 100 million population, GISAID on a country-by-country basis. (as of 27 May 2021) has determined that including Brazil (50 ppmp), India Our results suggest that researchers in the CST lag per strain ranges from 1 day (11 ppmp), Indonesia (6 ppmp), Nigeria United Kingdom are the fastest, logging to over 1 year. We have also calculated (4 ppmp) and Pakistan (1 ppmp), only sequences in a median time of 16 days, the median CST lag for each country. the United States (1,497 ppmp) and Japan which is not only more than 5 times as Examining the median CST lag values for (297 ppmp) have sequenced over 100 ppmp. fast as the upload times of sequences countries that have sequenced over 1,000 Cumulatively, African, Asian and South originating from industrial countries SARS-CoV-2 genomes, we note that the American countries have sequenced only such as Japan or Canada, but also 18 CST lag from the United Kingdom is the 14, 21 and 49 ppmp whereas this number times as fast as that of Qatar, among shortest (namely, 16 days for over 417,000 is 1,198, 948 and 607 ppmp for European, the countries that have sequenced genomes). This is almost a week faster than North American and Oceania countries over 1,000 genomes. the CST lags of 25 and 26 days for over (Supplementary Table 2). As of now, GISAID is the largest 590,000 and 498,000 genomes in the rest of Several reasons may explain the delay open-access portal, hosting the genome Europe and the United States, respectively. in sequence submissions to GISAID. sequences and related epidemiological For Canada, the CST lag is over five times Submission times are based on (i) the and clinical data of more than 1.7 million as long: 88 days for over 44,000 genomes. time taken from sample collection from SARS-CoV-2 strains. Thanks to ongoing Among the countries of Oceania, the CST a patient to RNA isolation in the lab and genomic surveillance using GISAID data, lag for New Zealand is 40 days for over its dispatch to the sequencing center and several new SARS-CoV-2 variants, such as 1,000 genomes, whereas for Australia it is (ii) the time from RNA sample arrival B.1.1.7 (Alpha; first identified in the United 51 days for over 17,000 genomes. In Asia, at the sequencing center to uploading Kingdom), B.1.351 (Beta; first identified the median CST lag is 72 days for over of the sequence to GISAID. Countries in South Africa), B.1.1.28 (Gamma; P.1, 89,000 genomes, with Singapore having the like the United Kingdom and Denmark first identified in Brazil), B.1.617.2 (Delta; shortest lag, 26 days for 2,405 genomes, with a short median CST lag have strong first identified in India), B.1.617.1 (Kappa; and Qatar the longest, 289 days for 2,298 public health systems, allowing efficient first identified in India), P.3 (Theta; first genomes. India’s median CST lag is 57 sample and metadata collection and identified in the Philippines), and B.1.427 days for 15,614 genomes whereas Japan, smoother coordination between the sample and B.1.429 (Epsilon; first identified in the which has sequenced the most genomes collection center, the RNA isolation lab United States), have been identified2–5. This in Asia, has taken 79 days for over 37,000 and the sequencing lab. Countries without information has been used to update public genomes. For South American countries, such a strong system are at a disadvantage health policies for the control of COVID-19 the median lag is 61 days for over 18,000 and may face additional logistical problems infections6,7. genomes, whereas countries in Africa in sample and metadata collection and Considering the benefits of genomic have taken 50 days for over 7,000 genomes shipping because of lockdown-related surveillance6,8, scientists have pressured (Supplementary Table 2). restrictions. Several countries might countries to increase their sequencing Coming to the rate of sequencing, have a shortage of labs that can handle capacity, and this has led to several top-performing countries Iceland, Australia, COVID-19 samples or might have an initiatives, such as COG-UK (United New Zealand and Denmark have sequenced overly centralized system wherein only Kingdom; https://www.cogconsortium. ~77%, 59%, 39% and 35% of their positive a few labs are authorized to handle such uk/), INSA-COG (India; https://pib.gov.in/ samples, respectively (Supplementary samples, causing a delay in sequencing PressReleseDetailm.aspx?PRID=1684782), Table 1). The United States and United and submission. A paucity of funds or NGS-SA (South Africa; http://www. Kingdom have sequenced over 400,000 restrictions on importing reagents and

1058 Nature Biotechnology | VOL 39 | September 2021 | 1055–1065 | www.nature.com/naturebiotechnology correspondence

525

Compact area

150

125

100

75

50 Collection to submission time lag (CSTlag)

25

Continent 0 Europe Italy Peru India Chile Israel Brazil Qatar Spain China

Asia Japan Tur key France Russia Ireland Poland Latvia Austria Mexico Croatia Iceland Finland Greece Estonia Norway Sweden Belgium Bulgaria Portugal Reunion Slovakia Slovenia Thailand Australia Canada Lithuania Denmark

North America Colombia Argentina Indonesia Singapore Philippines Hong Kong Bangladesh Netherlands South Africa South Korea South America Luxembourg New Zealand New United States Czech Republic Czech Oceania United Kingdom Africa United Arab Emirates United Arab

0

5,000

10,000

15,000

20,000

(27 May 2021) (27 May 25,000

30,000

Genomes submitted to GISAID Compact area 500,000

Fig. 1 | Violin plot illustrating the CST lag values for the 54 countries that have sequenced over 1,000 genomes. The box plot inside the violin plot depicts the median CST lag per country. Outlier CST lag entries are not shown. Country name is color-coded according to continent. We have also graphed the relative distribution of the number of genome sequences submitted by each country as a bar plot. equipment would also add to the delay. The listed above are known to have been true in particularly sensitive to the issue of virulent use of older sequencing technologies that India, for instance, and are being resolved10,11. strains being named after their countries. are low throughput and more expensive Sometimes, even after rapid sequencing, The WHO initiative of renaming variants per sample would complicate matters genomes may not be promptly uploaded with Greek letters may help resolve this further. to GISAID, and there may be several issue5. Finally, in many countries, significant Most of the countries with a short CST reasons for that. First, the importance of bureaucratization or political interference lag are industrialized nations that are likely genomic surveillance may not have been at various steps, from sample collection to have strong linkages between the clinical well understood, especially in the early to uploading sequences to GISAID, can and scientific establishments, although Japan months of the pandemic. Second, there add to the delay. In several countries— and Canada are outliers. This is not always may be a wish to withhold information, including the United States—where most of so for other countries. Some of the countries to publish or patent first. Although there the testing and sampling is carried out by with a longer CST lag have a less developed is a general understanding that scientists private diagnostic labs, there is no financial public health system. They might also have do not publish work based on others’ data incentive to share data, and the labs there had to establish new collaborations and unless the latter are acknowledged or have may prefer to discard samples rather than institutional arrangements to help deal with already published on them, there have been pay for their storage or shipping. Although the pandemic. All of this would have taken breaches of this professional norm, leading one does not know the extent of various time, which would have affected work on the to some hesitation in sharing unpublished problems in each country, clearly far ground. Some of the possible causes for delay data12. Third, several governments may be more samples have been tested than have

Nature Biotechnology | VOL 39 | September 2021 | 1055–1065 | www.nature.com/naturebiotechnology 1059 correspondence

11. Agrawal, A. Nature 594, 9 (2021). been sequenced; and far more samples have Kishan Kalia, Gayatri Saberwal and 12. Van Noorden, R. Nature 590, 195–196 (2021). been sequenced than are represented in Gaurav Sharma ✉ 13. Ascoli, C. A. Nat. Biotechnol. 39, 274–275 (2021). GISAID. Institute of and Applied Biotechnology In countries with a longer CST lag, the (IBAB), Bengaluru, India. Acknowledgements ✉ new variants may have enough time to e-mail: [email protected] G. Sharma is supported by funding from the 13 establish themselves across a region if Department of Science and Technology-INSPIRE quick tracking, tracing and actions to stop Published online: 10 August 2021 (DST-INSPIRE) program, Government of India. transmission are not undertaken. Therefore, https://doi.org/10.1038/s41587-021-01040-0 This work was partially supported by the Department this issue must receive urgent attention and of Electronics, IT, BT, and S&T of the Government References of Karnataka, India. The views expressed in this letter are bottlenecks that prevent a lower CST lag those of the authors, and not necessarily those of either 1. Lo, S. W. & Jamrozy, D. Nat. Rev. Microbiol. 18, 478 (2020). must be addressed. 2. Anonymous. Lancet 397, 445 (2021). funding agency or any other institution. Overall, an effective genomic surveillance 3. Burki, T. Lancet 397, 462 (2021). system requires not only sequencing a 4. Abdool Karim, S. S. & de Oliveira, T. N. Engl. J. Med. 384, major fraction of SARS-CoV-2 strains from 1866–1868 (2021). Competing interests 5. WHO. WHO Activities https://www.who.int/en/activities/ The authors declare no competing interests. COVID-19 patients, but also rapid genome tracking-SARS-CoV-2-variants/ (2021). submission to open access platforms like 6. Cyranoski, D. Nature 589, 337–338 (2021). GISAID. This will enable researchers across 7. Grubaugh, N. D., Hodcrof, E. B., Fauver, J. R., Phelan, A. L. & Additional information Cevik, M. Cell 184, 1127–1132 (2021). Supplementary information The online version the globe to track the evolved variants 8. Anonymous. Nat. Biotechnol. 39, 527 (2021). contains supplementary material available at https://doi. and their mutations, and 9. Geoghegan, J. L. et al. Emerg. Infect. Dis. 27, 1317–1322 (2021). org/10.1038/s41587-021-01040-0. biological consequences, which will provide 10. Ofce of the Principal Scientifc Adviser to the Government of India. Data access and enhancing research collaborations Peer review information Nature Biotechnology thanks the crucial inputs for appropriate and effective of urgent importance. https://static.psa.gov.in/psa-prod/psa_ anonymous reviewers for their contribution to the peer public health policies ❐ custom_fles/Data%20access_PSA%20website.pdf (2021). review of this work.

The need for new test verifcation and regulatory support for innovative diagnostics

To the Editor — The pressing need to in the RADx initiative, we gained a unique leadership and regulatory experts to address the COVID-19 pandemic led to the perspective on the frenetic pace of test convey clear regulatory guidelines and creation of novel large-scale cooperative development, performance testing and processes to developers, who often had programs among the US government, regulatory assessments in response to the little experience in this area. RADx also including the US National Institutes of COVID-19 pandemic. Consistent with an provided test developers access to a network Health (NIH) and the US Food and Drug editorial recently published here in Nature of experienced entrepreneurs to help them Administration (FDA); academia; and Biotechnology1, our experience suggests that avoid common mistakes that could hinder private industry. A case in point is the NIH’s RADx provides a high-reward approach for innovation, evaluation and deployment of Rapid Acceleration of Diagnostics (RADx), a relatively small investment. Furthermore, COVID diagnostics. the diagnostic testing arm of Operation we believe the RADx model could be The ACME POCT has evaluated over Warp Speed — a public–private partnership broadly established and applied beyond the 80 different diagnostic technologies for to fast-track SARS-CoV-2 vaccines, current pandemic as a foundational resource detection of SARS-CoV-2 infection. These diagnostics and therapeutics. As the name to improve the developmental pathway for products, which typically detect either suggests, the goal of RADx is to accelerate novel laboratory and point-of-care tests for SARS-CoV-2 antigens or SARS-CoV-2 RNA, development, verification, validation, FDA all varieties of diseases. originated from a spectrum of applicants Emergency Use Authorization (EUA) and For its RADx infrastructure, the NIH ranging from well-established companies deployment of diagnostic tests to detect leveraged its Point-of-Care Technology to small teams from academic laboratories. infection with SARS-CoV-2. Availability of Research Network (POCTRN), of which our For test verification at the ACME POCT, fast, accurate and inexpensive testing is a key center, the Atlanta Center for Microsystems each diagnostic technology was put component of efforts stemming from the Engineering Point-of-Care Technologies through a battery of different evaluations to pandemic. RADx was created in response (ACME POCT), serves as the national determine preliminary analytical and clinical to our nation’s declared state of emergency RADx test verification hub. In this role, performance (for example, limits of detection, that caused a sharp increased demand for and in collaboration with the FDA, we sensitivity, specificity and cross-reactivity), testing in traditional clinical laboratories, provided independent and impartial repeatability and usability. Whenever resulting in a nationwide shortage. This verification data on the performance of possible, tests were cross-compared with cascade of events motivated swift innovation COVID-19 diagnostic tests developed by one another to provide the NIH and RADx of new technologies using alternative test private companies or academic inventors. leadership the most objective performance materials and methods, and even alternative RADx and ACME POCT also created data on which to base their decisions to biospecimen sample types. As investigators a close working partnership with FDA fund test validation, further manufacturing,

1060 Nature Biotechnology | VOL 39 | September 2021 | 1055–1065 | www.nature.com/naturebiotechnology