LCR-BLAST—A New Modification of BLAST to Search for Similar Low Complexity Regions in Protein Sequences

Abstract

Low Complexity Regions (LCRs) are fragments of protein sequences that are characterized by a small diversity in amino acid composition. LCRs could play important roles in protein functions or they could be relevant to protein structure. However, for many years, low complexity regions were ignored by the scientific community which resulted in lack of algorithms and tools that could be used to analyze this specific type of protein sequences. Recently, researchers became interested in the so-called dark proteome and studies on such kind of proteins revealed that a vast amount of them include LCRs. Therefore, there is an urgent need to adapt existing methods or develop new ones that could be useful for analysis of LCRs, especially in the context of their functional roles in protein sequences. In this paper, we present LCR-BLAST which is a new modification of BLAST designed to search for similarities among LCRs. This modification consists of the following elements: turning on short sequence parameters, turning off compositional based statistics, applying our own version of identity scoring matrix and replacing E-value with mean-score statistics. In order to evaluate the performance of our new modification, we compare the number of similar pairs found by LCR-BLAST with performance of a standard BLAST tool and of BLAST with a specific set of parameters for compositionally biased and short sequences. We show that our new method provides a robust and balanced solution for searching for similarities among LCRs.