Selecting sequencing reads based on kmer counts

[tboothby]

Forgive my ignorance but is there a established method for selecting sequencing reads based on kmer count?

e.g., I want to parse only sequencing reads that contain kmer sequences with between 50-70X coverage in my dataset.

[westerman]

Interesting question. I have never done this but my first inclination would be to use Bedtools. Do an genomeCoverageBed with a '-bg' switch followed by a short Perl/Sed script to select columns with 50-70 in them. Then an intersectBed.

[krobison]

There are a number of tools for k-mer counting out there; khmer may have examples of precisely what you are looking for as that group does a lot of binning & normalization of reads based on k-mer frequencies