Bioinformatics Supergroup

The next Bioinformatics supergroup will be this coming Monday, April 7, at 4 pm in Ekeley W165/166. We are continuing the tradition of pairing talks that highlight bioinformatics methods (including new methods developed here at CU) with biological results using those methods. This Monday, Massimo Buvoli, in Leslie Leinwand's lab in MCDB, will talk about myosin mutations and computational and experimental approaches to understanding the effects of those mutations, and Greg Caporaso, in Larry Hunter's lab at CU Denver, will talk about approaches to find covariation and compensatory mutations, including applications to myosin.

Why you should attend: Myosin is an important molecule both for its role in muscle development and diseases such as cardiomyopathy, and has also been a model system for biophysics experiments for decades, providing many key insights into macromolecular assembly. Mutations in myosin are very important for medical reasons: however, mutations that we know cause disease in humans are often the wild-type in other species, raising the question of how these other species can get away with changes that cause us problems. The hypothesis is that other positions in the sequence have to change at the same time to compensate for these mutations, but figuring out which mutations compensate is an extremely challenging statistical problem. Greg's talk introduces several different methods for detecting covarying residues, some exploiting the phylogenetic tree. One might expect that inferring the tree structure and using it to help infer which mutations are important would give substantial improvements in accuracy, but, as Greg will describe, he found that, surprisingly, very simple methods that ignore the tree structure often produce better results. One innovative aspect of this work was using interactions between residues at successive turns of the regular alpha helix structure as a positive control for residues that are not adjacent in the sequence but that should interact with one another physically in the molecule, and therefore statistically during evolution. The workflow for studying mutations that Massimo, Greg, and Micah Hamady and Sandra Smit (in my lab) have put together should be useful for answering a lot of questions related to coevolution, including new ways of identifying protein-protein interactions: again, this represents a really productive interaction between bench and computational scientists.