[Biojava-l] Can anyone help me with MarkovModels?

[Schreiber, Mark]

Hi -
 
It sounds somewhat like some of you emmission or transition distributions do not have weights set. You can test this by setting a small nullmodel weight in the ModelTrainer before training. This will make sure none of the values is NaN.
 
If that doesn't work try posting a stack trace to the list.
 
- Mark
 

	-----Original Message----- 
	From: Oscar Lopes [mailto:oscar at gauss.inesc.pt] 
	Sent: Wed 23/04/2003 6:21 a.m. 
	To: biojava-l at biojava.org 
	Cc: 
	Subject: [Biojava-l] Can anyone help me with MarkovModels?
	
	

	Hi!
	
	I'm working with HMMs in ORF/Gene finding. A few days ago I defined a HMM
	relatively big (900 transitions and a lot of states). After
	initialization, I tried to apply Viterbi algorithm to a sequence, but I
	kept getting null pointer exceptions inside biojava classes. I also tried
	to apply
	Forward ans Backward algorithms, but I only got NaN's. Some of the states
	only emmit one symbol, so I defined different alphabets for these states,
	but instead of null pointer exception, I've got Illegal symbol exception.
	
	I wonder if there is a problem with HMM classes, since this kind of
	situation shouldn't occur. Is there a better solution, as some of the
	states emmit only one symbol?
	
	I'm using biojava version from CVS, a DNA alphabet, and I'm building the
	HMM as explained in the tutorial using SimpleEmissionState,
	SimpleMarkovModel, etc. If necessary, I can send some of my files.
	
	I would appreciate any help, since this work is for my degree thesis and
	I'm getting nervous :P
	
	Thanks in advance.
	
	Oscar Lopes
	
	--
	Oscar Manuel Baptista Lopes - oscar at gauss.inesc.pt
	Knowledge Discovery & BioInformatics - http://algos.inesc-id.pt/kdbio/
	Departamento Engenharia Informatica - http://www.dei.ist.utl.pt/
	Instituto Superior Tcnico - http://www.ist.utl.pt/
	
	_______________________________________________
	Biojava-l mailing list  -  Biojava-l at biojava.org
	http://biojava.org/mailman/listinfo/biojava-l
	


=======================================================================
Attention: The information contained in this message and/or attachments
from AgResearch Limited is intended only for the persons or entities
to which it is addressed and may contain confidential and/or privileged
material. Any review, retransmission, dissemination or other use of, or
taking of any action in reliance upon, this information by persons or
entities other than the intended recipients is prohibited by AgResearch
Limited. If you have received this message in error, please notify the
sender immediately.
=======================================================================