The Affiliation for Computing Equipment (ACM) awarded its first ACM Gordon Bell Particular Prize for Excessive Efficiency Computing-Based mostly COVID-19 Analysis to a multi-institution analysis crew that included the U.S. Division of Power’s (DOE) Argonne Nationwide Laboratory.
The crew was singled out for its work, ”AI-Pushed Multiscale Simulations Illuminate Mechanisms of SARS-CoV-2 Spike Dynamics,” which shines a lightweight on how the SARS-CoV-2 virus infiltrates the human immune system, setting off a viral chain response all through the physique. The award was introduced Nov. 19 at SC20, the Worldwide Convention for Excessive-Efficiency Computing, Networking, Storage, and Evaluation, held nearly this yr.
We’re excited to have received this prestigious award. The entire level is to push the boundaries of what we will do with AI. The flexibility to scale such an enormous set of simulations and use AI to drive some elements was key to this work.”
Arvind Ramanathan, Computational Biologist and Co-Principal Investigator, Argonne Nationwide Laboratory
Supporting a big collaboration of analysis organizations and scientific disciplines, Argonne exploring the usage of synthetic intelligence and high-performance computing sources to review, in nice element, the advanced dynamics of the spike protein, one of many key proteins within the SARS-CoV-2 virus.
The crew, comprised of almost 30 researchers throughout 10 organizations, is attempting to grasp how that protein binds to and interacts with one of many first factors of contact with the human cell, the ACE2-receptor protein. That binding begins a cascade of occasions that finally lets the viral and human cell membranes fuse, permitting the SARS-CoV-2 virus to enter and infect the host.
Proteins aren’t static, they’ve a variety of feelings that span a number of lengths- and timescales and it is not all the time understood which motions are necessary, notes Arvind Ramanathan, an Argonne computational biologist and co-principal investigator on the undertaking. To grasp and simulate these actions requires an enormous quantity of knowledge and computing sources.
Creating an affordable simulation of the spike protein alone can create an enormous system consisting of roughly 1.eight million atoms and simulations can encompass huge datasets that tax the sources of even the biggest supercomputers. To be able to make that information extra accessible for interpretation, the crew developed a machine studying technique that may summarize massive volumes of knowledge.
“One of many key issues that this technique allowed us to do was to find out what was fascinating, what was necessary, even these issues that weren’t apparent to the human eye,” stated Ramanathan. ”So, while you look deeper utilizing the simulations, you begin seeing vital adjustments within the protein construction, which informed us one thing about how the spike protein opens up such that it might work together with the ACE2 receptor.”
As the scale of the methods they have been engaged on grew, the crew confronted challenges of scaling all the information to run fluidly on at this time’s largest and finest supercomputing methods, in addition to their key parts.
As a result of lots of the machine studying fashions they have been coaching on these massive simulations wanted to be effectively scaled to be used on supercomputers, they partnered with NVIDIA, a pacesetter in GPU and synthetic intelligence design, to successfully run the fashions on Summit, on the DOE’s Oak Ridge Nationwide Laboratory.
The crew additionally utilized lots of the high U.S. supercomputers, together with Theta at Argonne; Frontera/Longhorn at Texas Superior Computing Middle; Comet at San Diego Supercomputing Middle; and Lassen at DOE’s Lawrence Livermore Nationwide Laboratory, to uncover alternative routes to deal with the deluge of knowledge.
“Given the complexity of the info, attempting to grasp the ACE2 receptor-spike interplay appeared virtually not possible at this scale,” Ramanathan confided. ”One of many issues that we clearly confirmed was that we might actuate a sampling of those dynamical configurations, pushing the concept that we might use AI to bridge these totally different scales.”
The info generated, up to now, is offering new insights into how the stalk area of the spike protein adjustments its total motions when it interacts with the ACE2 receptor, he stated. Ultimately, these sorts of insights derived from the extremely conjoined mixture of machine studying and simulation will assist facilitate antibody or vaccine discoveries.