Enquiry about the parameter "-m"

[cyr20040123]

Dear Dr. RUAN,

Thank you for providing this wonderful assembler.

I am wondering about the meaning of the parameter "-m" which is called "min match" in the documentation. To my understanding, when its value is 200, it means two bins can be aligned when the total length of k-mers shared between these two bins is not smaller than 200. Am I correct? If my understanding is correct, does it mean that when the sequencing error rate is higher, this value should be set smaller so that bins will be more easily to be paired/aligned?

Thank you!

[cyr20040123]

Also i have some minor questions about the pairwise binned sequence alignment. The default gap penalty is 7. It means gap extension penalty is 7 and there is no gap open penalty in Wtdbg2, am i right? Also since the penalty can be set, i would like to know what is the score gained for bin matching in Smith-Waterman algorithm? Thank you!

[cyr20040123]

By the way, is it possible to speed up the pairwise binned sequence alignment using some SIMD instructions like the Smith-Waterman in wtpoa-cns? Thank you very much for your time!

[ruanjue]

1, min_match: yes, you are right.
2, --dp-penalty-gap and --dp-penalty-var are not calculated in the same way with SW, they are just used to stop the tracing in sparse matrix when score < 0.
3, wtpoa-cns implemented an adaptive-banded-SIMD POA, but I am not satisfied with its SIMD, so I wrote another program bsalign to improve it. For the BIN alignment, it is not easy to use SIMD in a sparse matrix, where only a few cells are calcuated.