Heterogeneous Supercomputer Advances Viral Research and Animal Health
Total Page:16
File Type:pdf, Size:1020Kb
Heterogeneous Supercomputer Advances Viral Research and Animal Health Atos supercomputer with Intel® processors helps the Pirbright Institute safeguard livestock and humans from the rising threat of viral diseases The Context The Challenge The Solution When a deadly virus emerges, scientists • Rising demand for computational Taking advantage of Intel® technologies must respond rapidly to characterize resources threatened to outstrip and Atos expertise, Pirbright implemented the virus, track its spread, and stop it from the capacity provided by Pirbright’s legacy an Atos HPC system with heterogeneous devastating livestock and possibly infecting mix of servers, clusters, and workstations. nodes based on the Intel® Xeon® processor humans. As a global leader in this work, • Pirbright sought a versatile system E7 and E5 families. the Pirbright Institute in the UK needs that could accelerate progress in areas flexible high-performance computing (HPC) such as genome assembly of complex Atos Life Sciences Center of Excellence resources that can handle a wide variety viruses and hosts, epidemiology studies (LSCoE) experts worked hand in hand of workloads. Pirbright deployed an Atos to monitor virus migration, with Pirbright to provide a bespoke compute supercomputer powered by Intel® Xeon® and the development of innovative platform to enable researchers increase processors. With a unified environment analytics tools. the efficiency of their processes and running its diverse applications, Pirbright ultimately ensure the control of viral diseases. enhances scientific productivity and helps policymakers respond effectively when a viral outbreak threatens. Located in Surrey, England, Pirbright is the UK’s flagship research center focused The results • ·By running its diverse workloads on giving the UK capabilities on a unified environment, Pirbright lowers to predict, detect, understand, and respond management overhead and avoids to economically important viral diseases the time and expense of moving vast of livestock. This work aids in tracking amounts of data among multiple systems. the global spread of deadly viruses • ·Scientists get rapid results, gaining insights and developing diagnostic tools, vaccines, that can help reduce the impact of viral The solution has been deployed under treatments, and policies that can control, disease, ensure food security, and improve the umbrella of the Atos HPC, AI and contain, and eliminate them. Pirbright’s the quality of life for animals and humans. Quantum Life Sciences Center of Excellence, breakthroughs also suggest breeding Whole virus genomes are sequenced the realization of Atos global strategy to bring strategies that can increase disease within 24 hours, enabling pinpoint advance computing technologies closer to resistance. Strategic funding is from tracking of a virus’ path and aiding life sciences projects with the goal the UK’s Biotechnology and Biological the development of diagnostic tools, of discovering fresh, innovative solutions. Sciences Research Council (BBSRC). vaccines, and treatments. https://www.pirbright.ac.uk/about-us/our- mission-values “In an emergency situation, we need to sequence and analyse larger data volumes within the disease containment envelope. The Atos platform and Intel® processors give us the speed to carry out that analysis quickly and without all the problems of bandwidth and moving huge volumes of data around. This allows us to respond faster to the disease.” Dr Bryan Charleston Director and CEO The Pirbright Institute Preserving Life, the Economy and the Food Supply Agriculture is a vital element of the UK’s such as avian influenza crosses to human Pirbright houses research programs, economy, putting food on the table populations, the impact is even broader. diagnostic labs, bio-containment labs, and paychecks in the pockets of nearly and computational scientists in the same one-half million people. Now, virus activity is on the rise, due facility. “This leads to rapid movement of to factors such as population movement, information from the field, where Viral diseases can be deadly to livestock global trade, and climate change. biological samples are gathered, as well as economically devastating “We’re seeing an unprecedented movement to our research laboratories,” says Charleston. to the nation. A single, persistent virus—foot of viruses globally,” says Dr Bryan Charleston, “It also allows our research outputs to move and mouth disease (FMD)—has an annual director and CEO of The Pirbright Institute. quickly to people who are developing global impact of up to USD 21 billion in animal “Viruses are moving in a much more and implementing strategies for the control production losses and vaccination costs. dynamic system than we’ve seen for many of diseases in the field. We are constantly Even where FMD is well controlled, outbreaks years. New strains are emerging, providing advice to governments around of the disease still cause losses of more and previously localized strains the world as well as to pharmaceutical than USD 1.5 billion each year 1. If a virus are threatening new geographic regions.” and diagnostic companies.” 02 “Purchasing a supercomputer is not like going to a supermarket and picking something off a shelf, especially when you have so many different user applications to support,” says Dr. Paolo Ribeca, head of Pirbright’s Center for Integrative Biology and Bioinformatics. “We did a lot of design and discussion with Atos to define the system we wanted. Now we have a centralized resource with heterogeneous nodes so we can work on all the classes of problems we have to face. With the Intel® processors, we have excellent performance. And for all practical purposes, we have virtually unlimited storage.” Dr. Paolo Ribeca Head Center for Integrative Biology and Bioinformatics The Pirbright Institute Real-Time Insight Powerful, Flexible System In addition to accelerating Pirbright’s scientific for Diverse Workloads applications, the new platform speeds What makes one virus particularly virulent? the development of innovative analytical What enables some animals to resist a viral As the demand for computational resources and computational methods —an area infection? How is a virus changing has risen, Pirbright began to outgrow the mix in which Ribeca and the Center at a genomic, cellular, molecular, of workstations, servers, and small clusters it are international leaders—that are being and population level as it spreads? had implemented over the previous decade. customized for the study of viruses. Storage was a particular problem. To answer these and other questions, Different groups had local storage resources, “We’re doing non-standard method Pirbright scientists integrate genomics, so collaborative projects and mixed development for looking at new viruses proteomics, transcriptomics, and other workloads often required transferring large that have not been characterized previously,” “omics” data, along with epidemiological, data volumes between aging systems. says Ribeca. “We’re studying the genomics phenotypical, and environmental information, of viral swarms and the evolutionary and advanced methods in bioinformatics, Pirbright’s leaders saw an opportunity landscapes of viruses with a complex statistics, and mathematical modeling. to improve scientific productivity population structure—we need to model These studies are carried out not just by modernizing their legacy infrastructure. the interaction between those viruses, on viruses themselves, but also on vectors They worked with Atos to design the host and, possibly, one or more vectors. (mosquitos and midges) and hosts (livestock). a unified environment that could increase That requires ingenuity, and the correct By combining these methods, molecular performance and throughput combination of algorithms and statistics epidemiologists can trace the movement for the institute’s workloads. into novel methods. Thanks to the Atos of a virus and determine how it is evolving. platform and Intel technologies, we can develop such methods much more quickly, Many workloads are computed and/or data Performance and Memory and test them without having to move intensive. Genomes must be sequenced with Intel® Technologies terabytes of data through the Internet.” multiple times, often using different tools The Atos system with Intel technologies and protocols. Many viruses of interest The new system combines primary nodes is also more reliable, manageable, are RNA viruses rather than DNA viruses. built around the Intel® Xeon® processor E5 and energy-efficient compared to the These RNA viruses mutate so rapidly family, with memory-intensive tasks such previous legacy environment, Charleston that understanding their properties often as genome assembly running on Intel® observes. He and Ribeca also plan to explore requires analysing a swarm of closely related Xeon® processors E7-4809 configured new Intel technologies that are highly genomes, raising the complexity with 1.5 TB of memory. The environment relevant to their workloads, including Intel® of the genomics challenges. Fast turnaround also includes a half-petabyte of disk storage. 3D XPoint™ technology, a revolutionary is essential. “With high-throughput advance in nonvolatile memory and storage, sequencing and enough computational The choice of Intel technologies Intel® Xeon Phi™ processors, and future- power, we can actually plot transmission was straightforward. “Nowadays, there generation Intel processors. routes and see the differences at a forensic