Combining secondary structure element alignment and profile-profile alignment for fold recognition

Abstract

One of the most intensely studied problems of bioinformatics is the prediction of a protein structure from an amino acid sequence. In fold recognition, one reduces this problem to assigning a protein of unknown structure to one of the known fold classes as defined in the SCOP or CATH classifications. Here, we combine two alignment methods, secondary structure element alignment and log average profile- profile alignment that have been proven to perform well on this task. Our results show that the combination yields remarkably better fold recognition accuracy on well- known benchmark sets obtained from the literature. Especially on a difficult set built by McGuffin and Jones this new approach significantly outperforms other recently proposed fold recognition methods.