Theory, Methods, and Applications of Coevolution in Protein Contact Prediction

Publication date: Available online 18 August 2015 Source:Advances in Protein Chemistry and Structural Biology Author(s): Jianzhu Ma, Sheng Wang Coevolution is a fundamental mechanism of evolution to maintain structural or functional constraints on proteins. The increasing availability of protein sequences allows the efficient detecting of those coevolutionary couplings between residues in proteins and addresses the long-standing challenge to accurately predict protein contacts from amino acid alone. A wide range of theoretical and computational methods have been invested in studying the coevolution of pairs of positions in multiple sequence alignments of protein families. Recent theoretical developments based on direct-coupling analysis have improved the accuracy of contact prediction substantially by disentangling the direct pairwise couplings from indirect correlations. To incorporate heterogeneous features, a variety of machine-learning-based methods have been proposed, and the recent advancement is the deep learning approach to progressively refine the prediction. To further increase the accuracy, several hybrid methods were developed to utilize the output from other prediction methods, to exploit the information from other related protein families, or to employ constraints information that implicitly derived from the feasible protein structure. Here, we review the main coevolution-based computational approaches for protein contact prediction, their theoretical ba...
Source: Advances in Protein Chemistry and Structural Biology - Category: Biochemistry Source Type: research