I need to do pairwise alignment of nucleotide sequences of length 15-200 nt. Preferably, it should be written in C and accept *char variables for the sequences.
You can find, among examples, a SSE-vectorized implementation of Smith-Waterman algorithm implemented on top of FastFlow parallel programming framework. In this second link you can find the related paper (“Efficient Smith-Waterman on multi-core with FastFlow). Hope it is useful.
Thanks Claudia. I could run a test to compare FF to other implementations, but I do not know yet.
I am going to do many SW alignments on pairs of DNA sequences. Most of these sequences are no longer than 30 nt (1 nucleotide/nt = 1 character). My current idea is to run an SSE-vectorized implementation on each pairwise SW alignment, and have an parallel algorithm that runs one such alignment at a time per core.
I think that your paper has the idea of running several cores per SW alignment for long DNA sequences. That is different to what I am looking for.