Exact protein structure classification using the maximum contact map overlap metric

Abstract

In this work we propose a new distance measure for comparing two protein structures based on their contact map representations. We show that our novel measure, which we refer to as the maximum contact map overlap (max-CMO) metric, satisfies all properties of a metric on the space of protein representations. Having a metric in that space allows to avoid pairwise comparisons on the entire database and thus to significantly accelerate exploring the protein space compared to non-metric spaces. We show on a gold-standard classification benchmark set of 6,759 and 67,609 proteins, resp., that our exact k-nearest neighbor scheme classifies up to 95% and 99% of queries correctly. Our k-NN classification thus provides a promising approach for the automatic classification of protein structures based on contact map overlap. © 2014 Springer International Publishing.