75% Accuracy in Protein Secondary Structure Prediction

Abstract

In this study we present an accurate secondary structure predictionprocedure by using an query and related sequences. The most novelaspect of our approach is its reliance on local pairwise alignmentof the sequence to be predicted with each related sequence ratherthan utilization of a multiple alignment. The residue-by-residueaccuracy of the method is 75% in three structural states after jack-knifetests. The gain in prediction accuracy compared with the existingtechniques, which are at best 72%, is achieved by secondary structurepropensities based on both local and long-range effects, utilizationof similar sequence information in the form of carefully selectedpairwise alignment fragments, and reliance on a large collectionof known protein primary structures. The method is especially appropriatefor large-scale sequence analysis of efforts such as genome characterization,where precise and significant multiple sequence alignments are notavailable or achievable.