FIU algorithm harnesses power of supercomputers

By Adrienne Sylver 

In the world of medicine, time matters ― particularly when you’re in a race to save lives. An algorithm developed by FIU researchers from the Knight Foundation School of Computing and Information Sciences (SCIS) could potentially shave weeks, or even months, off the time necessary to crunch large amounts of complicated data that scientists use to create new vaccines and medications, to develop better treatments for cancer and Alzheimer’s disease, and to make inroads in the area of personalized medicine. The algorithm may be used for research in other fields as well. 

“Many of the instruments used by scientists have become more efficient in producing large data sets,” said Fahad Saeed, associate professor at SCIS. “But analyzing and processing the data has never been fast or efficient. We knew that if we could compress the data and share it across very large machines, we could speed up the process.”  

Saeed and his graduate students and postdoctoral fellows have been working on harnessing the power of high-performance supercomputers for a decade. In 2020, The National Institutes of Health awarded a grant to FIU, with Saeed as the principal investigator. The goal was to develop an algorithm to help scientists who use mass spectrometry to study proteomics ― proteins in a cell, tissue or organism.   

The information is vital in their work to comprehend and treat diseases. Saeed cited the need for a vaccine at the beginning of the COVID-19 pandemic as an example of the importance of having this sort of algorithm. While supercomputers have the ability to perform quadrillions of calculations every second, the high-performance framework for proteomics data analysis did not yet exist. While scientists produced several effective vaccines quickly, in the future and with the new algorithm and the use of supercomputers, the process could be streamlined. 

“Some people might say that running their data through a traditional algorithm and waiting six weeks or two months for the result is okay,” Saeed said. “But that is wasting taxpayer dollars, wasting human resources, slowing progress. It is human life. We argue that time is of the utmost importance.”   

With access to some of the largest supercomputers (think the size of a football stadium) in the United States, the FIU researchers have developed a computational solution. The result of their groundbreaking work, “High-Performance Computing Framework for Tera-scale Database Search of Mass Spectrometry Data,” has been published in the journal Nature Computational Science. The paper was authored by Muhammad Haseeb, doctoral student, and Saeed. See the full article here.  

“The publication places FIU in the forefront of this field,” Saeed said. “It shows that at FIU we have the human intellectual capital and the resources to produce groundbreaking work.” 

Because the code is open source, the algorithm (called HiCOPS) is available for free on the Saeed lab website for anyone interested. In the future, Saeed believes algorithms such as his, used in combination with advanced imaging technology, will also be used to help physicians better diagnose people with mental health issues, which currently rely heavily on observational data.