chrisquince / STRONG

Strain Resolution ON Graphs
MIT License
46 stars 9 forks source link

Pipeline always fails first time #43

Closed chrisquince closed 5 years ago

chrisquince commented 5 years ago

Pipeline seems to fail on first run but then run through when restarted at least on Test data.

Sebastien-Raguideau commented 5 years ago

Could you expand on what step it fails ? I just relaunched one right now on the strongtest vm, It is going to assemble.

chrisquince commented 5 years ago

This time it was rpsblast:

[Thu Aug 1 21:53:27 2019] Error in rule rpsblast_on_folder: jobid: 106 output: annotation/group1/group1_Rpsblast_cogs.tsv log: annotation/group1/group1_Rpsblast.log (check log file(s) for error message) shell:

    /mnt/gpfs/Hackathon/FMTMeren/STRONG/COG_pipe/scripts/Rpsblast_loop.sh annotation/group1/temp_splits 8 /home/sebr/seb/Database/rpsblast_cog_db/Cog &>log
    mv annotation/group1/temp_splits/Complete_Rpsblast_cogs.tsv annotation/group1/group1_Rpsblast_cogs.tsv

Job failed, going on with independent jobs.

chrisquince commented 5 years ago

Hmm... well Rpsblast_loop.sh does not exist anymore?

Sebastien-Raguideau commented 5 years ago

Hi Chris, I did some refactoring yesterday. I did remove Rpsblast_loop.sh and it's now handled by snakemake. I think it's better to handle interruptions. I am not sure where the issue you observed comes from, I'll try to replicate it on new Test instance but since I did some refactoring, it may not show up anymore.

chrisquince commented 5 years ago

OK Seb can you also add a flag in the config file to not construct the phylogenetic tree? Or we use a reduced database for testing. That is the slowest step of the Test pipeline.

Sebastien-Raguideau commented 5 years ago

Hi Chris, I'll do that quickly and maybe add some comment to the config file. Also, BayesPaths.py still throw errors on test data, could you have a look at it?

Sebastien-Raguideau commented 5 years ago

OK Seb can you also add a flag in the config file to not construct the phylogenetic tree? Or we use a reduced database for testing. That is the slowest step of the Test pipeline.

Just done it, the config file has an example : maganalysis: execution : 0