Open Sebastien-Raguideau opened 5 years ago
What version of BayesAGraphSVA are you using, can you make sure it is up to date with?
git submodule foreach git pull origin master
It seems to run, so it was just an outdated version of BayesAGraphSVA. This is cool. I'll relaunch from scratch and if everything works I propose we merge tomorrow.
It failed again. Though only after running for quite some time. It is the same bin and the log are still at : /mnt/gpfs/Hackathon/SyntheticCommunity32/Seb_Debug2/bayespaths/Logs/Bin_merged_1_simplif.log
File "/mnt/gpfs/Hackathon/STRONG/BayesAGraphSVA/AssemblyPath/AssemblyPathSVAS.py", line 1617, in sampleMargPath selectIn = self.prng.choice(np.arange(NC), 1, p=tempProb) File "mtrand.pyx", line 1144, in mtrand.RandomState.choice ValueError: probabilities contain NaN
At that point I think this is a BayesAGraphSVA issue more than a pipeline issue.
I agree this looks like a BayesAGraphSVA issue and I will debug it.
One thought though is do we want to reengineer the pipeline so that it does not break if one BayesAGraphSVA run fails but continues with the rest of the pipeline? We can discuss that this afternoon.
Are we skyping? That is easy to do, it's a snakemake option, we can add it permanently .
The Test dataset is handy but I can't debug merging event with it. So I've been using SyntheticCommunity32 also. Not sure if this is the new pipeline or an issue with bayesgraph. I'm going to start rerunning current pipeline with results from june and check for discrepancy.
If you have time, Chris, can you have a quick look? Here are the logs : /mnt/gpfs/Hackathon/SyntheticCommunity32/Seb_Debug2/bayespaths/Logs/Bin_merged_1_simplif.log