News

Seminar

Five Heirarchical Levels of Sequence-Structure Correlations in Proteins

Chris Bystroff
Rensselaer
Department of Biology

November 8, 2005
Sage 3510 - 4:00 p.m. to 5:00 p.m.
Refreshments at 3:30 p.m.

Abstract:

Expert classification of protein structures hints at the underlying physical process of protein folding and serves as a guide for a computational model for folding. Proteins are made up of conserved short sequence-structure motifs, which are patterns of amino acids that correlate with structure. These motifs are arranged in such a way along the chain that suggests a grammatical model, and are arranged in space in ways that suggest higher-order motifs.

In this talk, I will review about eight years of work on statistical modeling of sequence-structure correlations in proteins, leading towards a set of five heirarchical models paralleling the physical process of folding. The output of each model is the input of the next. Using the analogy of language models -- the first model condenses characters into words; the second, words into phrases; the third, phrases into associated pairs of phrases; the fourth, pairs of phrases into packed assemblies; and the fifth, packed assemblies into proteins. Molecular simulations based on statistical models will also be discussed.

Bio:

Hosted by: Jeff Trinkle (x8291)

Last updated: October 18, 2005