Thanks!! It is working, I found most of the missing alignments now, but what is the meaning of setting word_size to 2?
I want to use blastp to find out all the alignment among two fasta file.
Here is my command:
blastp -query "A.fasta" -subject "B.fasta" -evalue 1 -outfmt 6 > output.txt
I have already set the e-value tolerance to be 1, but there are still some alignments missing, is there anyway to force blastp to produce all the possible alignments?
1 answer
BLAST is a heuristic, it is not guaranteed to find all alignments. To increase sensitivity, try to decrease word_size (default is 3, try 2) and threshold parameters. Better, use FASTA or SSearch (Smith-Waterman) to get all local alignments which takes a lot of time. Oh, and set the evalue to 100, just in case ;)
@Michael Dondrup Thanks!! It is working, I found most of the missing alignments now, but what is the meaning of setting word_size to 2?
This explains it quite ok: http://en.wikipedia.org/wiki/BLAST#Process, In short, a smaller word size will make blast more sensitive and slower.
Log in to answer this question.
Btw, how do you know, the alignments exist in the first place?