This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Fastest Way To Search For Perfect Matches Only In Blast Or Blat

Hi, I have a numerous 100-mer sequences (let's say billions). What I am going to do is to query these sequences to entire human genome to find "perfect matches" only.

I first tried to do this using BLAST+ (megablast). I constructed blastdb and index using hg19. I gave below options to benefit from allowing perfect matches only 

blastn -query myseqs.fa -db hg19 -use_index true -index_name hg19_index -word_size 100 -outfmt "7 qacc qstart qend sacc sstart send sstrand" -max_target_seqs 5 -num_threads 4

Here, I gave word size '100' to achieve my goal. It does retrieve only perfect matches. But the problem is the speed, which is about a million queries per an hour. Well, someone can say this is fast enough, but I want it to be faster!

On the other hand, I could use BLAT instead of BLAST, which is generally accepted as a faster tool. I also constructed my local BLAT server (gfServer and gfClient), but I am not sure how to control BLAT parameters to get only perfect matches.

So, what would be the fastest way to retrieve perfect matches in BLAST/BLAT?

sequence alignment blat blast+
• 6,846 views • link
written by Sangwoo Kim • Sangwoo Kim

1 answer

Answering to my own question. I just solved this problem by using "bwa fastmap". Hope this was helpful to anybody with the same problem.

• 35 views • link
written by Sangwoo Kim • Sangwoo Kim

fastmap still does more than your need. In principle, we can have something several times faster than fastmap for your task.

• 39 views • link
written by lh3 lh3

Could you give me a few examples? I tried SSAHA but it was much slower than fastmap.

• 1 views • link
written by Sangwoo Kim • Sangwoo Kim

No, ssaha2 won't do. You need a new but very simple aligner to align a read in full length only.

• 1 views • link
written by lh3 lh3

Log in to answer this question.

More posts like this