Inferring protein function by local surface matching and similarity assessment: A Bayesian Markov Chain Monte Carlo approach based on geometric computation

Prof. Jie Liang
Department of Bioengineering, University of Illinois
Chicago

Room 610, CBI, New Life Science Building, PKU
2:00 PM, Thursday, 25 August,2005

Abstract:

Inferring biological roles of proteins and classifying them by their functions are challenging tasks, as global protein sequence and structure similarities are often unreliable for functional inference. Protein plays its role by interacting with other molecules, and local binding surfaces contain direct useful information. We study functional local surfaces of proteins that involve only a small number of key residues dispersed in diverse regions of the primary sequences. We develop methods for automatic identification of surface patterns and motifs in sequence, spatial arrangement, and spatial orientation that are likely to be biologically important. To identify locally similar binding surfaces and to assess their biological similarity, scoring matrix such as Pam and Blosum are not suitable, because residues on protein functional surfaces experience different selection pressure than residues in folding core. We develop methods for estimating replacement rates of residues based on a continuous time Markov model using Bayesian Markov chain Monte Carlo. Combined with geometrically computed libraries of millions of binding surfaces, we show our method can infer protein functions from structures with accuracy, and how to predict functional roles of proteins from structures of unknown biological roles that have only hypothetical sequence homologs.