Students Turn Summer Research into SC17 Research Poster
November 7, 2017
Instead of going on vacation last summer, recent computer science graduates Tom Corcoran and Rafael Zamora went to work at Lawrence Berkeley National Laboratory. They were invited to stay on at the lab because of their advanced work at the intersection of protein classification and machine learning. Their project also got them an invitation to present their work at the 2017 Supercomputing Conference in Denver.
At the conference, to be held November 12-17, the two will present a poster describing their work. First held in 1988, the conference is the leading international conference on high performance computing and is expected to draw 12,000 attendees from around the world. Corcoran and Zamora will join other authors of student research posters in presenting their research during a special reception on Tuesday, November 14.
Theirs is one of 28 posters that were accepted by the conference and just one of seven made by undergraduates. Their poster on “A novel feature-preserving spatial mapping for deep learning classification of RAS structures" will be part of the ACM Student Research Competition and vie for an award and an invitation to present in a national competition.
Corcoran and Zamora, along with their professor, Xinlian Liu from Hood College in Maryland, worked with lab scientist Silvia Crivelli to develop deep learning techniques to study Ras proteins, which play a critical a role in cellular proliferation. More than 30 percent of all human cancers, including 95 percent of pancreatic cancers and 45 percent of colorectal cancers, are driven by mutations of the RAS genes, which create mutant forms of the Ras proteins. These variants are ‘switched on,’ which may cause cancer cells to reproduce uncontrollably. If the Ras proteins and the various Ras mutations can be fully characterized, that could help pharmaceutical firms design drugs that would specifically dock with the mutated proteins to fight the cancer.
Currently the process to find the structure and interactions of a single protein requires a series of time consuming and expensive experiments. Computational science researchers are working on new methodologies to find the structure and predict the interactions of the protein without going through the mess of physical experimentation. However, these technologies are still far from perfect and create models of different quality. How to choose those biologically meaningful models is a challenging task.
Corcoran and Zamora are developing a new method that will greatly improve the results of this technology. Their software will take the frequently flawed protein models generated by these methods and determine how accurate or inaccurate they are. The results of the accuracy are placed on a scale of 0 percent to 100 percent accuracy. This will allow researchers to determine how reliable a protein model is and weed out any inaccurate data. The team hopes to apply their software to cancer research.
“This opens a whole new path for deep learning of structural data, for example, those produced in molecular dynamic simulations,” said Prof. Liu, Corcoran and Zamora’s professor at Hood College. “This won't be possible without guidance and encouragement from our Lab mentor Dr. Crivelli and the summer visiting opportunities. I am very proud of them and hopefully, their success will inspire more students from small colleges like Hood College to follow their path.”
Corcoran and Zamora started off in the lab as interns on what was just going to be a summer long project. However, their work was so promising that they were employed by the lab to continue their work with Silvia Crivelli.
“They were working at the level of seasoned graduate students and producing results of high quality,” said Crivelli, “so Prof. Liu and I decided that they were ready to submit to SC17. I’m sure their poster will attract lots of visitors and that many will be as impressed at these young researchers as we are.”
In December the team will be returning to Colorado to present a similar poster and an oral presentation on Dec. 8 at the 15th Annual Rocky Mountain Bioinformatics Conference, a conference hosted by the International Society for Computational Biology. Representatives from universities, industrial enterprises and other laboratories around the world will be attending the conference.
Editor's note: This article was written by Albany High School intern Cleighton Roberts.
About Computing Sciences at Berkeley Lab
The Lawrence Berkeley National Laboratory (Berkeley Lab) Computing Sciences organization provides the computing and networking resources and expertise critical to advancing the Department of Energy's research missions: developing new energy sources, improving energy efficiency, developing new materials and increasing our understanding of ourselves, our world and our universe.
ESnet, the Energy Sciences Network, provides the high-bandwidth, reliable connections that link scientists at 40 DOE research sites to each other and to experimental facilities and supercomputing centers around the country. The National Energy Research Scientific Computing Center (NERSC) powers the discoveries of 6,000 scientists at national laboratories and universities, including those at Berkeley Lab's Computational Research Division (CRD). CRD conducts research and development in mathematical modeling and simulation, algorithm design, data storage, management and analysis, computer system architecture and high-performance software implementation. NERSC and ESnet are DOE Office of Science User Facilities.
Lawrence Berkeley National Laboratory addresses the world's most urgent scientific challenges by advancing sustainable energy, protecting human health, creating new materials, and revealing the origin and fate of the universe. Founded in 1931, Berkeley Lab's scientific expertise has been recognized with 13 Nobel prizes. The University of California manages Berkeley Lab for the DOE’s Office of Science.
DOE’s Office of Science is the single largest supporter of basic research in the physical sciences in the United States, and is working to address some of the most pressing challenges of our time. For more information, please visit science.energy.gov.