drop reads containing specific set of kmers

[yannickwurm]

Hey @lh3, this looks great.

But we do low-coverage (5x) sequencing of many (non-human) individuals - where removing rare kmers is a bad idea. So our ideal approach is to combine all data into a big dataset (500-1000x coverage total), use that to identify bad kmers, dump those kmers to a file. Then go through each individual low-coverage dataset to eliminate the list of bad kmers. Can you add an option to bfc that can help with this last step? Or is it already hidden somewhere?

Cheers,
Yannick

[lh3]

500X-1000X total coverage is too much for bfc to handle. You could consider KMC2, though I don't how long it will take. You may also consider to ask @jts and Thomas Kean from Sanger. They are/were doing similar things.

[yannickwurm]

ok understand - thanks