Identification Of The Protein Superfamilies Based On The Distribution Of Different Amino Acids

Although protein families in the same protein superfamily have low sequence identities, their structures and function features suggest that they have a common evolutionary origin. The usual methods for identifying the protein superfamilies are to search for the common patterns of the proteins in the same superfamily by sequence alignments, and these patterns have been considered as the core to decide the structure and function of proteins. Other methods based on the analysis of physicochemical properties of the amino acids also have been proposed. Such as, the patterns of hydrophilicity and hydrophobicity are most important properties for recognizing protein superfamilies.At first, the distributions of the amino acids with different hydrophobicity and hydrophilicity along the amino acids sequence are taken as the parameters of the sources of diversity. The four different superfamilies in the same structure class are predicted by using the least increments of diversity algorithm. The results show that the overall prediction accuracies of differentsuperfamilies are 83.1% and 80.7% for all-α class, both 80.9% for all-β class, 88.6% and88.0% for α+β class, 69.3% and 67.6% for α/β class by self-consistency test andjack-knife test, respectively.After that, by considering the effect of the three special amino acids (proline, glycine and cysteine) with the significant contribution on protein structures, the numbers of these three amino acids are also taken as the parts of the parameters of the sources of diversity. For the same sets of...