Mutualism and machines: theory and software for genome analysis: Bioengineering Department Seminar, Professor Ian Holmes

Seminar: BioE Seminars: Grad | September 11 | 12-1 p.m. | 106 Stanley Hall

 Ian Holmes, Associate Professor, Bioengineering, UC Berkeley

 Bioengineering (BioE)

My lab develops software for genome informatics, from basic algorithms to user-friendly web tools. As I’ll describe, this connects us with many different kinds of biological research: 1) In one example, we’ve used state machines to develop a generalization of matrix algebra to probability distributions over sequences, allowing for vectors of infinite dimension. Theoretically, this establishes a fundamental connection between two classic algorithms of bioinformatics (Felsenstein’s pruning recursion and Sankoff’s multidimensional alignment); practically, we’ve used it to measure the rate of evolutionary drift, to reconstruct ancient viruses (and understand modern ones), and to develop basecalling tools for nanopore sequencing. 2) In other mathematical work, we’ve developed and extended “neutral drift” models for community ecology, proving that Hubbell’s unified neutral theory of biodiversity is related to the hierarchical Dirichlet process used in machine learning, and developing new clustering algorithms to identify mutualism vs neutrality. 3) On the software engineering side, we have developed JBrowse (the most popular browser for non-model genomics) and Apollo (the first Google Docs-like tool for collaboratively editing genome annotations), which have over 30,000 monthly active users. Even user interfaces need theory, and (while previewing new features of these tools) I’ll talk about how our genome browser design is influenced by Engeström's social object theory, a virtual sense of place, and prototyping ideas from the game industry., 510-642-5833