kmer size and alignments

Hi all -

Would anyone like to briefly discuss the idea behind kmer and alignment quality? I have searched around but am still unclear how changing the size affects the alignment.



Hey Jarrod,

I have played around with changing the ksize (default is 7) and I haven’t really seen a change in alignment quality. It seems that the ksize has less to do with finding the quality, and more to do about the speed . Here are some recommendation Pat gave me a while ago based on some simulations.

Yeah, the method is pretty insensitive to kmer size. If you check out my PLoS ONE aligner paper you can get a sense of this. A lot of the data are buried in the SOM. If the templates get shorter, then you generally need to decrease the kmer size.