An Approach for a Substitution Matrix Based on Protein Blocks and Physicochemical Properties of Amino Acids through PCA

Abstract

Amino acid substitution matrices are essential tools for protein sequence analysis, homology sequence search in protein databases and multiple sequence alignment. The PAM matrix was the first widely used amino acid substitution matrix. The BLOSUM series then succeeded the PAM matrix. Most substitution matrixes were developed by using the statistical frequency of substitution between each amino acid at blocks representing groups of protein families or related proteins. However, substitution of amino acids is based on the similarity of physiochemical properties of each amino acid. In this study, a new approach was used to obtain major physiochemical properties in multiple sequence alignment. Frequency of amino acid substitution in multiple sequence alignment database and selected attributes of amino acids in physiochemical properties database were merged. This merged data showed the major physiochemical properties through principle components analysis. Using factor analysis, these four principle components were interpreted as flexibility of electronic movement, polarity, negative charge and structural flexibility. Applying these four components, BAPS was constructed and validated for accuracy. When comparing receiver operated characteristic (ROC50) values, BAPS scored slightly lower than BLOSUM and PAM. However, when evaluating for accuracy by comparing results from multiple sequence alignment with the structural alignment results of two test data sets with known three-dimensional structure in the homologous structure alignment database, the result of the test for BAPS was comparatively equivalent or better than results for prior matrices including PAM, Gonnet, Identity and Genetic code matrix.

Analyses on clustering of the conserved residues at protein-RNA interfaces and its application in binding site identification

“…As the sequence entropy is needed in this work, the complexes where the protein has enough homologous sequences to calculate the sequence conservation will be remained. Multiple sequence alignments (MSA) were carried out by ClustalW [12,13] against the UniRef90 database [14] with default parameters and Gonnet substitution matrix [15] for all protein chains in 182 complexes. ClustalW, developed by Thompson, improves the sensitivity of progressive multiple sequence alignments through sequence weighting, position-specific gap penalties and weight matrix choice [12,13].…”