RecNW: A fast pairwise aligner for targeted sequencing

MotivationTargeted sequencing aims at in-depth analysis of specific genomic loci through high-throughput sequencing for applications such as resequencing or CRISPR gene editing. These applications require exact pairwise alignment algorithms to fully characterize large amounts of reads by comparison to the targeted locus, or reference. Optimal solutions to this alignment problem are provided by classic implementations of the global and semi-global versions of Needleman-Wunsch algorithms, but they remain computationally expensive due to their quadratic complexity in time and space.\n\nImplementationIn this paper we present RecNW, an open source C++ exact aligner packaged for Python that implements the semi-global version of the Needleman-Wunsch algorithm with affine gap penalty. RecNW utilizes low complexity of targeted sequencing libraries by aligning only unique reads, and recurrently using blocs of the alignment matrix between reads based on their similarities. Through this, RecNW performs exact alignment on average more than four times faster than gold standard comparable software.\n\nSoftwarehttps://github.com/AYahi/recNW