Hi, I am wondering if anyone has compared Smalt vs Blast. Two ways to use Smith-Waterman to map reads to a reference genome.
I recently got a hold of GPU-Blast and was wondering if it might go the same speed. If so, would it have more optimal alignments? The blast2sam.pl script would help me get the right output (forgetting about paired-end for now*), so the question boils down to 1) which is faster and 2) which has better matches. If anyone takes me up on the challenge, I think you could compare with what my computer currently has: 36 CPU threads vs 1536 GPU cores.
* followup question: is there a way to go about this while retaining paired-end-edness?