-
Notifications
You must be signed in to change notification settings - Fork 42
Running stochhmm
Paul Lott edited this page Jul 29, 2013
·
23 revisions
StochHMM implements standard HMM, HMM with duration (for viterbi algorithm) and in the future will implement hidden semi-Markov model architectures and algorithms. It grants researchers the power to integrate additional datasets in their HMM to improve predictions. Finally, it adapts HMM algorithms to provide stochastic decoding giving researchers the ability to explore and rank sub-optimal predictions.
To run the StochHMM command-line application you'll need a model file and a sequence.
##Command-line options:
###Required command-line options and files
-model <model file> import model file
-seq <sequence file> import sequence file in fasta format
###Non-stochastic Decoding: Different algorithms available for decoding
-viterbi performs viterbi traceback
-posterior Calculates posterior probabilities
If no output options are supplied, this will return the posterior scores
for all of the states.
-threshold <score>: Return only the States with a GFF_DESC, if they are
greater than or equal to the threshold amount.
-nbest <number of paths> performs n-best viterbi algorithm
###Stochastic Decoding:
-stochastic <Type>
Types:
forward performs stochastic traceback using forward algorithm
viterbi performs stochastic traceback using modified-viterbi algorithm
posterior performs stochastic traceback using posterior algorithm
-rep <number of tracebacks to sample>
###Output options:
-gff prints path in GFF format
-path prints state path according to state number
-label prints state path as labels
-hits prints hit table for multiple tracebacks for all states at each position
A couple example model files have been provided.
3_16_Eddy.hmm - GC rich model from Problem 3.16 of
"Problems and Solutions in Biological Sequence Analysis. M. Borodovsky and S. Ekisheva. Cambridge
Press, UK (2006)"
Dice.hmm - Dishonest Casino Dice model from pg 65 of
"Biological Sequence Analysis: Probabilistic models of proteins and nucleic acids. R Durbin, S.
Eddy, A. Krogh, and G. Mitchison. Cambridge Press, UK (1998)"
GC-skew.hmm - SkewR model for predicting R-loop forming regions in the human genome. See
"Ginno,P.A. et al. (2012) R-loop formation is a distinctive characteristic of unmethylated human
CpG island promoters. Mol. Cell, 45, 814–825."
$ stochhmm -model ../examples/Dice.hmm -seq ../examples/Dice.fa -viterbi -label
#Score: -539.062
Eddy Dice TRACK_NAME:TRACK1 StochHMM FAIR 1 48 . + .
Eddy Dice TRACK_NAME:TRACK1 StochHMM LOADED 49 66 . + .
Eddy Dice TRACK_NAME:TRACK1 StochHMM FAIR 67 78 . + .
Eddy Dice TRACK_NAME:TRACK1 StochHMM LOADED 79 112 . + .
Eddy Dice TRACK_NAME:TRACK1 StochHMM FAIR 113 179 . + .
Eddy Dice TRACK_NAME:TRACK1 StochHMM LOADED 180 192 . + .
Eddy Dice TRACK_NAME:TRACK1 StochHMM FAIR 193 270 . + .
Eddy Dice TRACK_NAME:TRACK1 StochHMM LOADED 271 289 . + .
Eddy Dice TRACK_NAME:TRACK1 StochHMM FAIR 290 300 . + .
$ stochhmm -model ../examples/3_16Eddy.hmm -seq ../example/3_16Eddy.fa -viterbi -gff
$ stochhmm -model ../examples/3_16Eddy.hmm -seq ../example/3_17Eddy.fa -posterior
$ stochhmm -model ../examples/Dice.hmm -seq ../examples/Dice.fa -stochastic viterbi -rep 10 -label
$ stochhmm -model ../examples/Dice.hmm -seq ../examples/Dice.fa -stochastic posterior -rep 10 -label