CCMS: A greedy approach to motif extraction

Abstract

Efficient and precise motif extraction is a central problem in the study of proteins functions and structures. This paper presents an efficient new geometric approach to the problem, based on the General Hough Transform. The approach is both an extension and a variation of the Secondary Structure Co-Occurrences algorithm by Cantoni et al. [1-2]. The goal is to provide an effective and efficient implementation, suitable for HPC. The most significant contribution of this paper is the introduction of a heuristic greedy variant of the algorithm, which is able to reduce computational time by two orders of magnitude. A secondary effect of the new version is the capability to cope with uncertainty in the geometric description of the secondary structures. © 2013 Springer-Verlag.