use blast+ to blast fine domains in proteins

Question

[lagartija]

Hi everyone.

I want to find particular functions in my genome so I tried this out : downloading a database from pfam (in my case http://pfam.xfam.org/family/PF00483#tabview=tab3) and then blast it against my genome.

It works if I download the unaligned sequences so I got the entire proteins. My problem is that I want a better specificity so I need the result of the blast against the domain only.

Do you know if it's possible using blastp to take into account dots or dashes ?

(I considered replacing the dots with a letter like X for example but the problem is that the e-value and bit-score might not be interpretable anymore)

Thank you very much, Sofia

[Sej Modha]

It sounds like the input is genomic sequences data in that case, are you employing BLASTX like approach to search against the specific protein database that you have downloaded? Have you explored other options such as HMMER?

[Mehmet]

if you have proteins you can use hmmer tool with pfam profile file to find out which protein domains your proteins belong to.

[lagartija]

I did use hmmer but it doesn't have the option to look through one family domain only. Then I get as result a lot of data I have to go though to analyse for only one function that I am interested in

[lieven.sterck]

ok, can you elaborate on your specific research question then? If you only input on pfam domain profile you will only get hits/info for that one, but we might have the wrong idea of what you want to achieve.

Answer 1

[lieven.sterck]

as Sej Modha and Mehmet have indicated: HMMER should be your tool of choice for looking (mapping) pfam domains. these are provided as HMMprofiles so blast will not work with them (getting the proteins where the domains are derived from is a valid work-around, but not necessary )

However, doing HMMER searches on a genome is quite resource intensive, so a for a blast alternative have a look for at the ProDOM data (also part of InterPro), which is a domain DB to be used with blast

[lagartija]

Thanks, that was what I was looking for :) ProDOM does have the option to enter a pfam accession number (after trying it I can say that it is way faster than my first method, but it can be complicated to do batch search because the results lacks the query ids. Anyway it is very useful !)