How to decrease computing time of reciprocal BLAST

Question

[biolab]

Hi, everyone,

I am using Reciprocal Best Hits (RBH) method to group orthologous proteins from diverse species. I have 50 eukaryotic species at hand, so should do 49x50 BLASTs. I found it needs much time, even setting evalue = 1e-5 and getting top hit.

Is there any way to improve my work? I learned some parallel technology is powerful, but don't know how to start. Any of your suggestions will be appreciated. THANK YOU very much!

[IsmailM]

What is the command that you have tried?

[biolab]

Hi IsmailM, The following is my command. I am sure BLASTP is time-consuming. I am trying UBLAST (a program in USEARCH), which is better. However, handling 50 species is still time-consuming. Any of your suggestions will be appreciated.

blastp -query qfile -db dbfile -evalue 1e-5 -max_target_seqs 1 -outfmt 6 > result 2>log

Answer 1

[Buffo]

What about DIAMOND? I think it is a very good alternative to blastp.

Answer 2

[IsmailM]

Blastp takes an argument (-num_threads) where you can specify the number of threads to use. Run the following

blastp -num_threads 8 -query qfile -db dbfile -evalue 1e-5 -max_target_seqs 1 -outfmt 6 >result 2>log

See the following for more information:

blastp -h

Or for even more information:

blastp -help

[biolab]

Hi IsmaiM, thank you for your suggestions.