Mathematics Colloquia and Seminars

Return to Colloquia & Seminar listing

Description

Bacteriophages are viruses that infect bacteria and shape microbial communities. This talk follows the phage genome as both biological information and physical material. As biological information, the genome can help identify which bacteria a phage infects when experimental host evidence is missing. As physical material, the genome must fit inside a small viral capsid, raising the question of how long DNA organizes under tight confinement.  The first part of the talk treats the phage genome as sequence data for predicting bacterial hosts. I will discuss how a genome language model can represent phage and bacterial genomes, and how these representations can be used to retrieve likely hosts. I will then describe a machine-learning approach that asks how much host information the sequence evidence supports. The same framework can make specific host predictions, return broader host lineages, and suggest candidate missing hosts for future testing.   The second part focuses on bacteriophage P4, whose DNA is often knotted after extraction. We model the packed genome as an ordered liquid crystal and ask how this internal organization shapes the knot patterns observed in experiments.  

Also on zoom https://ucdavis.zoom.us/j/98969645841