kmc - 3.2.4+dfsg-1+b1 main

The kmc software is designed for counting k-mers (sequences of
consecutive k symbols) in a set of reads. K-mer counting is
important for many bioinformatics applications, e.g. developing de Bruijn
graph assemblers.
.
Building de Bruijn graphs is a commonly used approach for genome
assembly with data from second-generation sequencing.
Unfortunately, sequencing errors (frequent in practice)
result in huge memory requirements for de Bruijn graphs, as well
as long build time. One of the popular approaches to handle this
problem is filtering the input reads in such a way that unique k-mers
(very likely obtained as a result of an error) are discarded.
.
Thus, KMC scans the raw reads and produces a compact representation
of all non-unique reads accompanied with number of their occurrences.
The algorithm implemented in KMC makes use mostly of disk space rather
than RAM, which allows one to use KMC even on rather typical personal
computers. When run on high-end servers (what is necessary for KMC
competitors) it outperforms them in both memory requirements and
speed of computation. The disk space necessary for computation is in
order of the size of input data (usually it is smaller).