In a standalone blast search, is there a parameter that could be specified to fix the number of mismatches allowed in an alignment. I went through the parameters and I could only find parameters to set score.

-q => Penalty for a nucleotide mismatch (blastn only) [Integer] default = -3

-r => Reward for a nucleotide match (blastn only) [Integer] default = 1

What do I do if I only require hits with a maximum allowed mismatch or gap of 2 per alignment.


1 Answer 1


There is no direct option like that but you can set percentage identity filter with -perc_identity

However, this is only for reporting. BLAST will still perform all the alignments. That is why I suggested you to use bowtie.

EDIT-1 (getting long for comment)

If you want a word type of search then use the -n option. But it always starts from the 5'end which is quite justified in case of miRNA because the seed region is in the 5'.

Setting -n to 0 means no mismatch in seed region. -l option defines the seed length; I would advise that you use something like 10. -e refers to maximum mismatches in the non seed region but it is not the number of mismatches; it means maximum tolerated sum of phred score of mismatches (fasta files are assigned a default phred score of 40 at all residues. So, for 2 mismatches you have to mention -e 80).


Yes. Increasing mismatch and gap penalties would make the search more stringent if you have already specified a cutoff on score.

  • $\begingroup$ got it.. gonna go for bowtie.. I guess it would be -v 2 for the maximum number of mismatches set to 2, and -e 0.01 for the quality score, --norc for the plus strand and which parameter would be the equivalent of the word match limit. $\endgroup$ Commented May 27, 2014 at 9:11
  • $\begingroup$ and one last question for future reference, do you think increasing the mismatch penalty from -3 (default value) to maybe -5 or -6 would serve this purpose? $\endgroup$ Commented May 27, 2014 at 9:15

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .