Authors: J.Y. Yu; W. Zhang
Addresses: Department of Automation, Shanghai Jiaotong University, Shanghai 200240, China ' Department of Automation, Shanghai Jiaotong University, Shanghai 200240, China
Abstract: In protein structure prediction, backbone fragment bias information can narrow down the conformational space of the whole polypeptide chain significantly. Unlike existing methods that use fragments as building blocks, the paper presents a probabilistic sampling approach for protein backbone torsion angles by modelling angular correlation of (Φ, ψ) with a directional statistics distribution. Given a protein sequence and secondary structure information, this method samples backbone fragments conformations by using a backtrack sampling algorithm for the hidden Markov model with multiple inputs and a single output. The proposed approach is applied to a fragment library, and some well-known structural motifs are sampled very well on the optimal path. Computational results show that the method can help to obtain native-like backbone fragments conformations.
Keywords: protein backbone fragments; conformational space sampling; backbone torsion angles; directional statistics distribution; data mining; bioinformatics; hidden Markov model; HMM; structural motifs; protein structure prediction; polypeptide chain; probabilistic sampling; modelling; protein sequences.
International Journal of Data Mining and Bioinformatics, 2013 Vol.7 No.2, pp.180 - 195
Received: 06 Nov 2010
Accepted: 05 Mar 2011
Published online: 29 Mar 2013 *