The latest episode of the This Week in HPC podcast features Jim Brase, who works with the Computing Directorate at Lawrence Livermore National Laboratory (LLNL). Intersect360 Research’s Addison Snell spoke to Brase to discuss LLNL’s approaches to fighting COVID-19 on multiple fronts and Brase’s own workflow changes as COVID-19 forces LLNL to adjust its day-to-day operations. We present this lightly edited transcript alongside the podcast audio.
Addison Snell: Hi everyone, thanks for listening to another episode of This Week in HPC with Intersect360 Research, distributed in partnership with HPCwire. I’m Addison Snell with Intersect360 Research, and this week in HPC, I’ve got a special episode where I’m joined by Jim Brase, who’s in the Computing Directorate at Lawrence Livermore National Laboratory. Jim, thanks very much for joining me this week in HPC.
Jim Brase: Hey, happy to be here.
Snell: I really wanted to talk to you because we’ve had a huge focus, of course, worldwide, on the COVID-19 (or novel coronavirus) pandemic, and you’ve been involved in a lot of work going on in the national lab efforts as we’re trying to dedicate supercomputing resources to help scientists, epidemiologists, pharmaceutical companies, and so on in the general research battle against COVID-19. To start, can you talk about some of the myriad ways that supercomputing can get deployed that help researchers in the battle against the virus’ spread?
Brase: Computing is used almost across the board in understanding and building models for how we can respond to COVID-19. It’s used in analyzing the structure of the virus with systems like x-ray crystallography or cryo-electron microscopy; it’s used in understanding the interaction between the virus and human cells using things like molecular dynamics simulations; it’s used in identifying targets on the viral proteins or in the co-complexes of the virus with human proteins. You go from there to computational design and screening of large sets of molecules, of antibodies, vaccines, all the way to the epidemiological models, flow models of ventilator systems used in hospitals. HPC really is applied across a large area and set of applications for this.
Snell: Lawrence Livermore is one of the premier supercomputing sites in the United States. Can you talk about how Livermore is involved specifically, and how you’re involved specifically – are you touching all of these types of applications, are you specializing in any particular area? What’s going on at Livermore?
Brase: Yeah, we have several specific projects that we have started up based on work that we’ve been doing in computational biology and predictive models in biology for many years. We have work specifically on analyzing the viral proteins and trying to understand what the best spots on those proteins would be for targeting with either small molecules or antibodies. So we’ll take things like the new crystal structures that are coming out, combine that with the sequence information that we have for the virus, and build high-resolution models of specific interaction sites and so on.
Then we have a project that we’ve actually been working on for quite some time, which comes out of a partnership we’ve developed over the last five years called ATOM, which stands for “Advancing Therapeutic Opportunities in Medicine.” It’s a collaboration across a number of [Department of Energy] labs with the National Cancer Institute, GlaxoSmithKline and the University of California at San Francisco. The collaboration is really aimed at building a platform for accelerated molecular design for rapid production of medicines. So we’re actually taking a lot of the tools that we’ve developed there with focus on cancer applications (because NCI is a partner) and repurposing those tools and using them to develop potential new antiviral molecules for this outbreak. So that’s one of the major projects we have.
We’re doing large-scale screening of molecules, we’re doing optimization of the molecular structures to enhance their safety and their pharmacokinetic properties. Then ultimately we’re feeding those to newly emerging experimental capabilities — some at the National Labs, some at partners like UCSF — to actually test those and validate which designs might actually be useful for application here. So that’s another area of work.
Finally, we have a very similar project that we’re doing in the computational design of antibodies. So these are large molecules, complexes of proteins, that interact with the virus and stop its operation and can do that very effectively and rapidly. So these are designs of much larger molecules than we’re doing for the small-molecule antiviral drugs, but ones which can be produced fairly rapidly and don’t have as many of the potential safety issues and therefore can be approved faster. So working on antibody treatments like this has been a particular area of focus for us as well.
We have a number of designs ready to go on this front that are being manufactured right now, and we’re hoping to get initial experimental feedback on those antibodies as soon as the next few weeks. We’re really excited about this, and hopefully we get good results out of this – at least results that we can feed back into the designs for the next round of work. We’re pretty optimistic about that approach.
Snell: This really strikes at the heart of what’s going on with regard to the urgency in this project. Oftentimes when people talk about scientific research, there’s a lot for the good of humanity and building our general scientific knowledge that has some future benefit that we might not fully understand. But now we’re talking about a project that’s real and present in our lives right now, and we look at daily trackers that tell us how many cases, how many lives lost, and people really want scientists to catch up as fast as we can. So what’s going on now, do we have early results coming out of the lab?
Brase: Yeah, we have designs that we’ve done, we have those starting experimental testing now, so the urgency on this is very high right now. As you said, we’ve been talking about this and working on this for a number of years, in various application areas. It’s been clear to us as we’ve looked at the possible threats that face us in these areas that a pandemic infectious disease like this is a likely thing. It’s happened before, it’s happening now, it’ll happen again. Being prepared to rapidly produce specifically designed therapeutics is a capability that we really strongly need in this country and in the world to be prepared for this. Viral diseases like this will continue to emerge and we need to have the tools in place to be able to respond rapidly.
Snell: And as we pointed out, you’ve got access to some of the most powerful supercomputing resources in the world with Sierra and other supercomputers. How subscribed are you with this kind of work right now? Are things getting reprioritized, are non-COVID-19 projects getting put on hold, where are we in supply versus demand of supercomputing power?
Brase: Lawrence Livermore and our sister DOE labs host the most powerful supercomputers in the world. We are absolutely prioritizing these systems to work on COVID-19. There’s no higher priority for our computing resources right now. I wouldn’t say we’ve stopped all the other research that we’re doing: we work on a lot of very important and high-priority national security problems at Livermore, a lot of leading-edge science problems, those are still continuing. But the teams that are working on COVID-19 response, whether it’s in specific therapeutic design, in trying to understand the function and structure of the virus – those are getting the top priority.
Snell: You made an excellent point with regard to the importance of these applications, not only for COVID-19 but also for other application areas, like, for example, national security. I talked to Carlo Cavazzoni of CINECA in Italy a couple of weeks ago on an episode of this podcast, and he was right in the epicenter of the Italian outbreak there that was so devastating. We are in northern California which is one of the original hotspots in the United States and it’s interesting to look at how it’s affecting us both professionally and personally. What’s considered essential when it comes to Lawrence Livermore National Lab? How are things operating now? How are you coping?
Brase: As far as the lab goes, we’ve been on minimal safe operations for several weeks now where the first priority is to maintain the safety of the lab and the people who are there. We’re continuing to run our computer systems, we’re continuing to have some minimal level of lab operations, particularly in areas that are related to the COVID-19 response, and there are a few specific national security priority projects that are continuing to operate at a fairly low rate but getting the important work that they’re doing done at a reasonable level. For me personally, being in the computing department at Livermore, I can pretty much work remotely as well as I can there most of the time. But just getting used to working completely through virtual meetings and so on has been a little bit of a learning experience, and we’re looking forward to getting back to whatever the new normal will look like sometime in the next weeks and months here.
Snell: And we’re all trying to figure out what that new normal is going to look like. It feels to me a lot like things are changing very dynamically on a daily basis — sometimes the world looks different by evening than it did first thing in the morning. How do you think this might evolve your work at Livermore over the coming weeks? I know this is also a priority at the entire DOE level, there’s a consortium evolved here. How do you see this evolving?
Brase: One thing that I hope we’ll see is a continuing prioritization of building up a sustainable capability for rapid response, having the open data and transparent research on how we design new molecules for medicines and so on. As you mentioned, this has also fostered some really new and unprecedented levels of cooperation on getting things like computing resources to groups all around the world that can take advantage of them. The DOE has really come together as part of this COVID-19 HPC Consortium to help with that. That’s something I hope is a sustainable result as well: seeing the levels of cooperation that we’re seeing between agencies, between labs and so on is great.
Snell: I’ve been speaking with Jim Brase, who’s managing multiple computing projects within the Computing Directorate at Lawrence Livermore National Laboratory. Jim, thank you very much for taking some time out of your very busy schedule to join me on the podcast today. I’ll let you get back to work, I’m very grateful for the important work that you’re helping to oversee.
Brase: My pleasure, Addison, good to talk to you.