Summary

The khmer software is a set of command-line tools built around a Python library designed for analysis of large DNA sequence collections. Functionality in khmer has primarily been motivated by scaling issues with (meta)genome and (meta)transcriptome assembly (Crusoe et al. 2015). khmer provides convenient access to several k-mer based operations on DNA sequence collections, such as abundance filtering, error trimming, assembly graph partitioning, and most notably, abundance normalization of reads (C. T. Brown et al. 2012) and streaming error trimming of reads (Zhang, Awad, and Brown 2015). All of these operations utilize khmer's implementation of two primary data structures, the Bloom filter and the Count-Min Sketch, for efficient probabalistic storage of k-mer presence or k-mer abundance, respectively (J. Pell et al. 2012; J. A. C.-K. Zhang Qingpeng AND Pell 2014).

Release version 2.1 of the khmer software includes several new features that extend its utility to a wider set of sequence processing and analysis problems. These include the following: support for variable-coverage trimming of sequence reads; support for k > 32 using the non-reversible hash function MurmurHash3 (https://github.com/aappleby/smhasher); a new optional Count-Min Sketch implementation providing increased storage efficiency; support for assembly directly from a k-mer graph; a script for computing a compact de Bruijn graph from a k-mer graph; and several examples of khmer's Python and C++ APIs for those interested in using and extending the library.

References