NextGenMap (NGM) is a flexible and fast read mapping program that is more than twice as fast as BWA, while achieving a mapping sensitivity similar to Stampy or Bowtie2. NextGenMap uses a memory efficient index structure (hash table) to store the positions of all 13-mers present in the reference genome. This index enables a quick identification of potential mapping regions for every read. Unlike other methods, NextGenMap dynamically determines for each read individually how many of the potential mapping regions have to be evaluated by a pairwise sequence alignment. Moreover, NextGenMap uses fast SIMD instructions (SSE) to accelerate the alignment calculations on the CPU. If available NextGenMap calculates the alignments on the GPU (using OpenCL/CUDA) resulting in a runtime reduction of another 20 - 50 %, depending on the underlying data set. Our results show that NextGenMap using only the CPU is at least twice as fast as BWA. Using the GPU for the alignment calculations increases the speedup to a factor of three. NextGenMap (GPU) even outperforms Bowtie2 by 10 - 50 % in terms of runtime. More importantly, the number of correctly mapped reads is similar to Stampy, one of the most sensitive methods available.

References in zbMATH (referenced in 1 article )

Showing result 1 of 1.
Sorted by year (citations)

  1. Keith, Jonathan M. (ed.): Bioinformatics. Volume I. Data, sequence analysis, and evolution (2017)