Seminar: David Burnstein

The language of microbial genomes and their mobile genetic elements

Unraveling the function of uncharacterized microbial genes is a fundamental challenge with tremendous discovery potential. Our group tackles this by integrating natural language processing approaches and comparative genomics. I will present two projects: In the first, we investigate how plasmids overcome bacterial defenses by encoding diverse anti-defense systems (e.g., anti-CRISPRs, SOS inhibitors, and anti-restriction genes) in their leading regions, which enter recipient cells first. In the second, I will discuss our exploration of language model applications to biological data, treating genomes as "sentences" and genes as "words" to reveal genomic grammar and gene semantics. Our work demonstrates that language-based approaches can reveal organizational modules, improving our understanding of microbial gene function and especially horizontal gene transfer mechanisms, opening new avenues for clinical and biotechnology development.