Closed genec1 closed 3 years ago
Hi @genec1 — Sorry you ran into this issue. The root cause looks to be:
05b2c11e443c: 2020-11-16 21:55:39,295 WARNING: toil.leader: 3/R/jobSC18zG: disk = '2G' if config.ci_test else r1_id.size + r2_id.size + 80530636800 # 75 G for STAR index and tmp files
05b2c11e443c: 2020-11-16 21:55:39,295 WARNING: toil.leader: 3/R/jobSC18zG: AttributeError: 'NoneType' object has no attribute 'size'
Given that you're running single samples,r2_id
is set to None
which is a pretty embarrassing bug. I tried to find this in the code but it does not exist in the latest version, which is when I noticed that you're running the quay.io Docker container. The container is unfortunately out of sync with the main codebase (the group changed CI after I moved on, which is what built/pushed the containers). So you'll either need to build your own Docker container or use the python (pip
) version of the code.
While I generally support using containers wherever possible, the python version is the preferred way to run this given the above issue. I also didn't write the Docker support for this pipeline and have noticed several other issues besides the one you ran into.
You will likely need to change the Toil version after pip
installing to: pip install toil[all]==3.12.0
as the team made several backwards-breaking changes after I left. Apologies that support for this is limited, but I graduated a while ago. Feel free to continue to open issues or let me know if you run into any problems and I'll try and get to them when I can.
Cheers, John
Thanks for letting me know that I'm on a deprecated version! I just tried running toil installed via pip as you recommend. Unfortunately I'm having issues with that too, but I'll post that as a separate issue.
My attempt at a TOIL run on a SRA-derived fastq file is dying. Some samples do run to completion, but many -- like this one -- fail.
Here is the full run. Any debugging assistance is appreciated.