Extract Unmapped Reads From Blast

Question

[Lorddoskias]

Is there a way to tell blast to output the unmapped reads into a separate file or at least the name of the unmapped sequences?

[Michael Schubert]

How do you map your reads with BLAST? To what? I don't understand your question.

Answer 1

[Torst]

I assume you have used blast to align some reads "reads.fa" onto a reference "ref.fa", and have a blast report in "out.bls" ?

% formatdb -i ref.fa -p F
% blastall -p blastn -i reads.fa -d ./ref.fa -o out.bls

As RM answered, you can (hackily) get the read IDs that got no hits. I have added extra code to ensure a clean list of IDs only into "nohits.ids":

% grep -B5 "***** No hits" out.bls | grep '^Query=' | sed 's/^Query= //' > nohits.ids

Now if you want to get back the sequences listed in nohits.ids from reads.fa, you can use this trick:

% formatdb -i reads.fa -p F
% fastacmd -d ./reads.fa -i nohits.ids -D 1 -o nohits.fa

Good luck!

Answer 2

[Rm]

If i understand your query correctly: This will extract sequences without hits (unmapped sequences) from blast (BLASTN 2.2.25+) output.

grep -B5 "***** No hits found *****" blast.output.txt | grep Query=