Blast sequence algorithm

[Niraks]

I've got a question regarding the BLASTn search algorithm that I can't quite grasp. Is it possible for two DNA sequences of length 100 may be >90% identical but have no similarity according to the BLASTn search and if so, how?

Is it to do with differences in alignment with the HSP or it's threshold values?

[krobison]

If they have that high an identity, I can't see how they wouldn't match for BLASTN even with it set to very high stringency. Are you sure everything is being run correctly and that they are really that similar?

[lh3]

If the 9 mismatches on a 100bp sequence happen to occur at the 11th, 22nd, 33rd,..., 99th positions, BLAST under the default configuration (11-mer) will miss it as there are no seed hits. This is pretty rare, though.