MIT Researchers Leverage Machine Learning for COVID Vaccine Research

Researchers at MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL) have published a paper describing a new combinatorial machine learning (ML) system that could both decrease the research time needed to develop a COVID-19 vaccine and make it more effective. 

The paper describes an open-source platform called OptiVax that leverages ML to select short strings of amino acids called peptides that could support high population coverage for a vaccine. Unlike whole-virus, DNA, and RNA vaccines, more than 100 of which are currently in development, peptide vaccine are developed around short peptide fragments of the type found in the target diseases. The vaccines use a synthetic version of the peptide created in a lab to engineer the induction of highly targeted immune responses.

Among other advantages, peptide vaccines are not burdened with large amounts of genetic information common to the other vaccine types, which aren't useful in developing resistance and can lead to unwanted immune responses and dangerous reactions. The OptiVax platform has the potential to reduce vaccine development costs because of its use of multiple predictive models, which minimize the risk of false positives, the researchers said.

Once the peptides are selected, OptiVax uses ML to ranks them according to their potential to elicit an immune response in a population, and selects those most likely to produce this response in the most people to maximize coverage.

When it comes to COVID-19, the MIT researchers aren't focused on the development of a pure peptide vaccine, but a complementary peptide that could be used to improve traditional vaccines by targeting the spike proteins that cover the virus. 

"We evaluated a common vaccine design based on the spike protein for COVID-19 that is currently in multiple clinical trials," said Ge Liu and Brandon Carter, CSAIL PhD students and lead authors on the paper, in an article on the MIT CSAIL Web site. "Based on our analysis, we developed an augmentation to improve its population coverage by adding peptides. If this works in animal models, the design could move to human clinical trials."

Next comes animal testing, followed by a decision about whether a clinical trial is warranted. Meanwhile, the researchers are working with a team at the National Institute of Health (NIH) to see if their methods can be used for risk prediction using data from COVID-19 patients.

In addition to Liu and Carter, the paper, "Computationally optimized SARS-CoV-2 MHC class I and II vaccine formulations predicted to target human haplotype distributions," was authored by MIT professor David Gifford, CSAIL post-doctoral student Siddhartha Jain, Trenton Bricken of Duke University, and Mathias Viard and Mary Carrington of Basic Science Program, Frederick National Laboratory for Cancer Research and the Ragon Institute of MGH, MIT, and Harvard.

About the Author

John K. Waters is the editor in chief of a number of sites, with a focus on high-end development, AI and future tech. He's been writing about cutting-edge technologies and culture of Silicon Valley for more than two decades, and he's written more than a dozen books. He also co-scripted the documentary film Silicon Valley: A 100 Year Renaissance, which aired on PBS.  He can be reached at