How to perform a BLAST of many sequences?

[Hel]

Hi all,

I have about 50,000 sequences that need a BLAST, is it possible to perform a BLAST search of they all in one step?

Thanks

[mastal]

How to perform a BLAST of many sequences?

Yes, you can do it using commandline blast+.

http://www.ncbi.nlm.nih.gov/books/NBK1763/

[Kaas]

I found this tutorial really helpful http://ged.msu.edu/angus/tutorials/unix-and-blast.html

[mastal]

How to perform a BLAST of many sequences?

It does seem like a helpful tutorial, however, it uses blastall, the old legacy blast, rather than blast+, and the commands/options for blast+ are
different.

[mike.t]

running the blast is trivial. You better think about what you're going to do with a blast report that contains 50,000 query sequences.

[Hel]

Thanks to all,
So it's better to use Blast+ than Blastall, right? Then I will try to find the Blast+ tutorial. About what to do with 50,000 query sequences...well, of course I've though about it. I think a blast search is the first thing to do, but I gladly accept suggestions

[mastal]

The best thing to do depends on what your 50,000 sequences are, whether they are raw reads or assembled contigs, for example.

If they are raw 454 reads, then the first thing to do would be to align them to a reference, with an alignment program - the Roche/454 software works quite well on 454 reads - or to do a de novo assembly. You could use an assembler like Mira, or Newbler, the 454 assembler.

[mike.t]

You'll probably have to write a script in a programming language such as Perl or Python to analyze the blast report and find the information you're looking for. You should consider this before you run the blast because how you parse the blast report might determine the parameters of the blast. For example, blast by default will report the first 200 hits from the database. If you're only interested in the first hit, then using the default parameters will result in a blast report much larger than you need. If you use BioPython to parse the report, you'll probably want to use xml output format. for such a large blast report it may be faster parsing to output the blast in tab separated format and avoid biopython and bioperl (also BioJava) altogether.

I recommend that you do a small blast with perhaps the first 100 sequences and then figure out how you're going to extract the info that you need. a blast of 50000 sequences is going to take a while so you probably don't want to do it twice...