[Biojava-l] Write and read profile HMM to disk

[Lachlan Coin]

Hi Tisanai,

I have also had problems with this approach.  From memory, the
problem is that the XML code for hmms does not support emission
probabilities properly, so when you read back a hmm you get a model with
the right transition probabilites, but the default uniform emission
probabilities (something like this at least - it was a while ago).  You
can confirm that this is the case by looking in the XML output.

Perhaps a better way to do this is via serialisation?

Lachlan


> Subject: RE: [Biojava-l] Write and read profile HMM to disk
> Date: Mon, 30 Sep 2002 10:46:57 +1200
> From: "Schreiber, Mark" <mark.schreiber@agresearch.co.nz>
> To: "Tisanai" <T.Krisanathamakul@dcs.shef.ac.uk>, <biojava-l@biojava.org>
> 
> Hi  -
> 
> Can you show us the code that you used to try and do this? I have
> succesfully read and written XML markov models so it can work (at least
> in some cases).
> 
> - Mark
> 
> 
> > -----Original Message-----
> > From: Tisanai [mailto:T.Krisanathamakul@dcs.shef.ac.uk] 
> > Sent: Monday, 30 September 2002 10:26 a.m.
> > To: biojava-l@biojava.org
> > Subject: [Biojava-l] Write and read profile HMM to disk
> > 
> > 
> > Hi
> >    I tried to save profile HMM and read it back via 
> > XmlMarkovModel but it didn't work. How can I cope with this problem?
> > 
> > Thank you
> > 
> > Tisanai
> > _______________________________________________
> > Biojava-l mailing list  -  Biojava-l@biojava.org 
> > http://biojava.org/mailman/listinfo/biojava-l
> >