2vcf.py exits with a non-zero exit code when executed by snakemake

[aryarm]

For some reason, the 2vcf.py script sometimes finishes with a non-zero exit code. It isn't clear what might be causing this. The usual warnings that I added are printed, but the warnings don't seem to be the cause of the non-zero exit code. No other information appears to explain why the exit code isn't 0. In fact, the VCF output looks correct.

This is a problem because the non-zero exit code causes Snakemake to remove the output after it is created.

[aryarm]

Sources online say that PyQt can sometimes cause scripts to silently finish with a non-zero exit code. So I might look into that...
So far, I've only been able to reproduce the problem when running it on an entire dataset with Snakemake. I get an exit code of 0 when running locally.

[aryarm]

Ok, new idea: it's because I didn't put matplotlib.use('Agg') at the top of the script

import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt

Update
wait - no it's not that, either
it seems like the command only fails in bash strict mode with set -o pipefail. That's why it wasn't failing when I ran it but it was failing when Snakemake ran it

[aryarm]

ok so it seems like the underlying issue is that 2vcf.py is not consuming all of its stdin, so zcat is erroring with exit code 141 and the pipefail is propagating that. Don't ask me how I figured that out. It took like 30 re-tries and a bunch of guess and check

so instead of piping the file, let's just try to read it from the python side
pandas should be able to handle gzipped files, anyway
I'll just need to tell it to ignore a bunch of the columns in the file, and apparently you can do that by passing a lambda function to usecols