Development of a method for Protein Remote Homologue Detection Using Secondary Structure Prediction and Solvent Accessibility Prediction.

Katsunori Uehara (0251137)


When we compare protein pairs, there are many protein pairs that share the similar fold but have no clear sequence similarity (remote homologues). To detect these remote homologues, we propose a new simple method which combines the standard sequence search method with secondary structure and solvent accessibility predictions. First sequence search method is performed normally to get the sequence similarity (E-value) between query and template sequences, next, secondary structure and solvent accessibility are predicted for both query and template aligned sequences by sequence search method. And then, Structure similarity between predicted query and template structures are evaluated using Zsocre. Final decision on whether they are remote homologous is made by the combined score considering both E-value of the standard sequence search method and the structure similarity. We used test using SCOP database as the correct standard to check this method’s performance. This study showed that the combination of E-value of sequence search method, secondary structure and/or solvent accessibility predictions improved BLAST/PSI-BLAST for detecting remote homologues.

Keywords: Protein、Homologue、sequence homology、secondary structure、solvent accessibility