Retrieve sequences that generated a HMM profile

Question

[frcamacho]

Hi All,

I have HMM profiles but the raw sequences or multiple sequence alignments are not available. I saw the tool hmmemit, but this randomly generates sequences that match a given profile HMM. I do not want random sequences. Is there a way to obtain sequences or multiple sequence alignment that generated the HMM profile? Additionally, I know skylign generates hmm logos, is there a way to obtain the consensus sequences from the logo? The other option would be to programmatically parse the HMM profile to try to generate the consensus sequence.

Any input would be appreciated!

Thanks in advance.

[Joe]

hmmemit is your only option as far as I'm aware. It doesn't randomly emit a sequence, it produces the most likely sequence given the hidden states of the Markov Model profile (a kind of consensus sequence).

Answer 1

[Sej Modha]

I do not think it is possible to get those sequences as HMM algorithm does not save them, however, you can use hmmemit as you mentioned. It is possible to tweak the hmmemit parameters to match the profile. Please refer to the HMMER documentation.

Answer 2

[Jean-Karim Heriche]

Unless the original sequences were saved somewhere, you can't get them back from the HMM alone. hmmemit -c will give you a consensus sequence.