This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Parse hmm(er) profiles in R

Hello everyone,

I am currently working with HMMER3 and created a bunch of profiles which I now want to use for another project.

The problem is, that I want to write the code for my project in R and can't find a method to load the profiles. Does anybody know a method or package to get the model out of the hmmbuild profile and store them in a way which can be used productively?

If there is no method available in R I could also live with Python, but I would prefer R.

Many thanks in advance.

hmm r parse hmmer profile
• 4,417 views • link
written by hivemind • hivemind

Another option I can think of is that with hmmsearch you can use different flags to save the results in a table. Which you could use R to parse and format the data how you intend to use it. 

 --tblout <f>     : save parseable table of per-sequence hits to file <f>
 --domtblout <f>  : save parseable table of per-domain hits to file <f>
 --pfamtblout <f> : save table of hits and domains to file, in Pfam format <f>
• 0 views • link
written by frcamacho frcamacho

1 answer

Hi,

thanks for the replies. I think I have to clarify my request: I am not trying to parse/load the results of hmmsearch (this should be no problem using SearchIO in Python). I want to load the hmmbuild model. Basically the values for the match-states, insertion-states and the transition-probabilities for each residue.

Is there anybody who might be able to tell me, if this is possible/reasonable or should I generate a new hmm profile with one of the hmm packages (like depmixS4)?

• 0 views • link
written by hivemind • hivemind

A profile HMM is basically a series of emission probabilities and a state transition matrix. hmmbuild output these in plain text and the format is described in section 8 of the HMMER manual. Not a ready-made solution but it can be used to build your own reader.

• 0 views • link
written by Jean-Karim Heriche Jean-Karim Heriche

Log in to answer this question.

More posts like this