Recent advances in improving the memory efficiency of the TRIBE MCL algorithm

Abstract

A fast and highly memory-efficient implementation of the TRIBE-MCL clustering algorithm is proposed to perform the classification of huge protein sequence data sets using an ordinary PC. Improvements compared to previous versions are achieved through adequately chosen data structures that facilitate the efficient handling of symmetric sparse matrices. The proposed algorithm was tested on huge synthetic protein sequence data sets. The validation process revealed that the proposed method extended the data size processable on a regular PC from previously reported 250 thousand to one million items. The algorithm needs 10-20% less time for processing the same data sizes than previous efficient Markov clustering algorithms, without losing anything from the partition quality. The proposed solution is open for further improvement via parallel data processing.