Cache-Efficient FM-Index Variants for Mapping of DNA Sequences

Abstract

Mapping reads to a reference genome is the first and very important step in genome analysis. One of the most used data structure for DNA read mapping is the FM-index. To this day many different variants of the FM-index exist. In the proposed paper we introduce two new variants of the FM-index suitable especially for DNA sequences. Proposed variants decrease the number of cache misses by efficiently interleaving auxiliary data structures of the FM-index thus increasing the searching speed. Experimental results have shown that the proposed variants are about two times faster than other variants, while having comparatively low memory requirements.Source code is available at https://github.com/xsitarcik/DNASeqMap.