July 16, 2020 — The University of Arizona, in partnership with research computing centers at Arizona State University and Northern Arizona University, is contributing research computing resources to a worldwide effort to advance COVID-19 research.
Arizona’s three state universities are participating in the national [email protected] project, which relies on volunteers’ idle computing power to run protein modeling computations that help researchers learn more about how to cure or treat certain diseases.
The project, based at Washington University in St. Louis, began in 2000 and focuses on cancer, infectious diseases and neurological diseases. The universities’ advanced supercomputers will be used during their downtime to contribute to the project, which is now largely focused on the COVID-19 pandemic.
The University of Arizona has three supercomputers, each with much more processing power and memory capacity than the standard desktop computer. The three computers are all connected, allowing them to speak to each other very quickly so researchers can do millions of small calculations – such as simulating 12 million galaxies over 400 million years.
Blake Joyce, assistant director of research computing at the University of Arizona, compares the machines to cars: If a desktop computer is a standard sedan, “a supercomputer isn’t one F1 race car; it’s more like having a fleet of cars.”
[email protected] provides its free software to volunteers or citizen scientists to download to their personal computers, allowing simulations of complex scientific processes to run in the background while computers are not in use. The project uses shared computing power at a massive scale to help solve grand challenges in biomedical research.
Researchers request computations via the [email protected] software, which assigns the computing tasks to computing providers. The computing providers send the results back through the software.
A number of NAU researchers are using the university’s supercomputer, Monsoon, to study COVID-19. Monsoon administrator Christopher Coffey said this additional approach offered an efficient way to use NAU’s high-performance computing capacity to the fullest. It’s also a natural partnership among the three public universities.
“For six years, we have collaborated on research computing projects whenever possible to help advance research in Arizona,” Coffey said. “This project combines the computing power at each of the universities to take on COVID-19 together and is another way NAU researchers are collaborating with our colleagues to help lead this fight.”
What is Folding?
The project seeks to understand how proteins – large, complex molecules that play an important role in how our bodies operate – “fold” to perform their biological functions. This helps researchers understand diseases that result from protein “misfolding” and identify novel ways to develop new drug therapies.
How proteins fold or misfold can help researchers understand what causes diseases like cancer, Alzheimer’s disease and diabetes. It might also provide insight into diseases such as COVID-19.
“Imagine if I told 100 people to fold a pipe cleaner,” said Joyce, who has a background in ecology, biology, genetics and bioinformatics. “They are going to fold it in 100 different ways because there’s an infinite number of combinations of how to take something that is straight and fold it. That’s what viruses and living things do with proteins. They make copies of themselves and fold them up in their own particular way.”
Using computational modeling, researchers can explore the mechanics of proteins of the virus and predict every possible way it might fold or physically change shape.
“In biology, shape is function,” Joyce said. “If you can disrupt that shape, the virus is inactive or can’t do its thing. If you disrupt any of the mechanisms that can damage us, you have a cure, or at least something you can treat. And that is what we’re after. It just takes a lot of computing to come up with every possible way to bend a pipe cleaner.”
By running computer simulations, researchers can see how the virus would interact with various compounds or drugs and narrow down which ones might work to interrupt one of the critical mechanisms the virus needs to survive.
[email protected] assigns pieces of a protein simulation to each computer and the results are returned to create an overall simulation. [email protected] computations for COVID-19 research seem to be most productive on the kind of computers found in facilities like Arizona’s research computing centers, making their contributions especially valuable.
Arizona’s Early Impact
Volunteers can track their contributions on the [email protected] website and combine their efforts as a team, receiving points for completing work assigned to them and even earn bonus points for work that is more computationally demanding or that might have a greater scientific priority. The point system helps [email protected] determine which machines in the project are quick and reliable.
The Arizona Research Computing team – which includes supercomputers across the three state universities – has risen quickly in the ranks, highlighting the universities’ powerful computing capabilities and the effectiveness of regional collaborations. As of early July, the team was ranked in the top 60 of nearly 250,000 teams, surpassing other teams that include Hewlett Packard, Cisco Systems, Apple and Google, as well as many other universities, industry and national or international contributors.
“Our systems really jumped up in points because we can process analyses reliably and quickly,” Joyce said.
The [email protected] project “investigates many research questions that require an enormous amount of computing, but this specific use for COVID-19 provides a unique opportunity, spurring many computing centers to participate in [email protected] for the first time,” says Gil Speyer, lead scientific software engineer for Arizona State University’s Research Computing center.
Source: University of Arizona