Table of Contents

Class: PairHMM /amd/bernoulli/1/home/abt_vin/georgi/hmm/0.7/ghmm//ghmmwrapper/ghmm.py

Pair HMMs with discrete emissions over multiple alphabets. Optional features: continuous values for transition classes

Base Classes   
HMM
Methods   
__init__
__str__
addEmissionDomains
checkEmissions
checkTransitions
logP
viterbi
viterbiPropagate
  __init__ 
__init__ (
        self,
        emissionDomains,
        distribution,
        cmodel,
        )

create a new PairHMM object (this should only be done using the factory: e.g model = PairHMMOpenXML(modelfile) ) @param emissionDomains: list of EmissionDomain objects @param distribution: (not used) inherited from HMM @param cmodel: a swig pointer on the underlying C structure

  __str__ 
__str__ ( self )

string representation (more for debuging) shows the contents of the C structure pmodel @return: string representation

  addEmissionDomains 
addEmissionDomains ( self,  emissionDomains )

add additional EmissionDomains that are not specified in the XML file. This is used to add information for the transition classes. @param emissionDomains: a list of EmissionDomain objects

  checkEmissions 
checkEmissions ( self,  eps=0.0000000000001 )

checks the sum of emission probabilities in all states @param eps: precision (if the sum is > 1 - eps it passes) @return: 1 if the emission of all states sum to one, 0 otherwise

  checkTransitions 
checkTransitions ( self,  eps=0.0000000000001 )

checks the sum of outgoing transition probabilities for all states @param eps: precision (if the sum is > 1 - eps it passes) @return: 1 if the transitions of all states sum to one, 0 otherwise

  logP 
logP (
        self,
        complexEmissionSequenceX,
        complexEmissionSequenceY,
        path,
        )

compute the log probability of two sequences X and Y and a path @param complexEmissionSequenceX: sequence X encoded as ComplexEmissionSequence @param EmissionSequenceEmissionSequenceY: sequence Y encoded as ComplexEmissionSequence @param path: the state path @return: log probability

  viterbi 
viterbi (
        self,
        complexEmissionSequenceX,
        complexEmissionSequenceY,
        )

run the naive implementation of the Viterbi algorithm and return the viterbi path and the log probability of the path @param complexEmissionSequenceX: sequence X encoded as ComplexEmissionSequence @param complexEmissionSequenceY: sequence Y encoded as ComplexEmissionSequence @return: (path, log_p)

  viterbiPropagate 
viterbiPropagate (
        self,
        complexEmissionSequenceX,
        complexEmissionSequenceY,
        startX=None,
        startY=None,
        stopX=None,
        stopY=None,
        startState=None,
        startLogp=None,
        stopState=None,
        stopLogp=None,
        )

run the linear space implementation of the Viterbi algorithm and return the viterbi path and the log probability of the path @param complexEmissionSequenceX: sequence X encoded as ComplexEmissionSequence @param complexEmissionSequenceY: sequence Y encoded as ComplexEmissionSequence Optional parameters to run the algorithm only on a segment: @param startX: start index in X @param startY: start index in Y @param stopX: stop index in X @param stopY: stop index in Y @param startState: start the path in this state @param stopState: path ends in this state @param startLogp: initialize the start state with this log probability @param stopLogp: if known this is the logp of the partial path @return: (path, log_p)


Table of Contents

This document was automatically generated on Fri Jan 20 14:56:59 2006 by HappyDoc version WORKING