starting HMM and HDP models

[PengNi]

Hi everyone, when using the trainModels program, how do I choose the needed HMM and HDP models as the starting models? Or, how to use the --in_template_hmm/--in_complement_hmm, --templateHDP/--complementHDP parameters?

[PengNi]

Hi @mitenjain ,

May you help me with this issue?

I used two .model files in /models folder to run the BuildModels.py script in _signalAlignnotebook , with Ecoli CCWGG R9 data. But I didn't get the expected results showed in the paper.

The training and testing positions were splited as described in the paper. And the command I ran is like this:

python ../signalAlign_notebook/experiment_scripts/BuildModels.py \
-r SRP098631/ecoli_k12.fasta \
-pcr SRP098631/ecoli/08_20_16_R9_WGA_Ecoli/MJ/downloads/pass/ \
-gen SRP098631/ecoli/08_05_16_R9_gEcoli_2D/DD/downloads/pass/ \ 
-o SRP098631/trainModels_Ecoli.CCWGG.try5 \
--positions SRP098631/ecoli_ccwgg_positions/cytosine.train.positions \ 
--positions SRP098631/ecoli_ccwgg_positions/cytosine.test.positions \ 
--motif SRP098631/ecoli_ccwgg_positions/cytosine.train.target \
--motif SRP098631/ecoli_ccwgg_positions/cytosine.test.target \
-x cytosine2 -j 36 --split 0.5 -a 30000 -i 30 \
--in_template_hmm ../signalAlign/models/testModelR9_acegit_template.model \
--in_complement_hmm ../signalAlign/models/testModelR9_acegit_complement.model \ 
-n 1000000 -e 1000000 --hdp_em 1 \
1>trainModels_Ecoli.CCWGG.try5.log \
2>trainModels_Ecoli.CCWGG.try5.err

So for the Ecoli CCWGG data, are the _testModelR9_acegittemplate.model and _testModelR9_acegitcomplement.model correct? Or there are other parameters wrong?

Best, Peng

[PengNi]

@mitenjain @ArtRand , Maybe I should move the issue to the _signalAlignnotebook repo?

Peng

[mitenjain]

Hi @PengNi ,

That looks right. I am pinging Art so that he can confirm and send appropriate info.

-Miten

[mitenjain]

Hi @PengNi ,

I have messaged Art who will respond to your query with specific details.

-Miten