Does anyone know how to build a local conserved domain database from multiple sequence alignment for running RPS-Blast locally?

Question

[BioPotato]

Does anyone know how to build a local conserved domain database from multiple sequence alignment for running RPS-Blast locally? What I need to do is aligning the protein sequence to those conserved domains to find the protein sequence with functional similarity. Can anyone help me? I really appreciate it.

Answer 1

[marchlerbauer]

You can use the psiblast application to convert mFASTA to a score-matrix/PSSM. You need to do a fake search:

psiblast -in_msa [my_msa] -db [some_db] -out_pssm [my_pssm.smp]

Note that the sequence database does not really play a role in this, so you would want to use a small/decoy database to avoid wasting time on an actual database search. You will have to generate a file containing the names of all the PSSMs you want to include in your RPS-BLAST search databse, e.g.:

ls *.smp > mysearchdb.pn

and then use the makeprofiledb application to build the RPS-BLAST search database, e.g.:

makeprofiledb -title mysearchdb -in mysearchdb.pn -out mysearchdb -scale 1 -threshold 11 -dbtype rps -index true

.. hope this helps!