Image courtesy of Flickr.
Researchers from Massachusetts Institute of Technology (MIT) have recently developed a machine learning method called OptiVax to evaluate the effectiveness of potential SARS-CoV-2 peptide vaccines. Such vaccines contain peptides, or short protein segments, that prime the immune system to respond to future viral infections.
OptiVax primarily scores a peptide on how well it activates T cells, a crucial member of the immune response. Specifically, it considers the match between the amino acid sequence of both the peptide and major histocompatibility complexes (MHCs), which are molecules that bind to the peptide and “present” it to appropriate T cells.
A major hurdle of designing a SARS-CoV-2 peptide vaccine is the diversity of human MHC genes, which confers varying binding affinities between MHCs and respective peptides. The MIT researchers looked towards analyzing the human leukocyte antigen (HLA) loci, a genetic marker that mediates such binding affinities, within MHCs. Using this strategy, they predict over ninety percent population coverage for their proposed SARS-CoV-2 peptide vaccines.
This study demonstrates the potential for machine learning systems to propel vaccine development, simultaneously decreasing the required research time and enabling researchers to navigate novel avenues for effective implementation.