Analysis files from wrong samples removed

[parlundin]

When restarting NGI_PIPELINE for a sample where previous analysis already been performed, the previous analysis files are removed from the analysis folders 01 to 07, unfortunately it looks like it in some cases removes the files for other samples as well.

[Galithil]

I pushed a bugfix for that, but I'm going to need to test it. Is there a way to try it without doing that in production ? reproducing the bug seems a little hard

[mariogiov]

Yeah this is weird. @parlundin do you have an example you can direct us to?

[Galithil]

patrik's words :

I believe it happend when I restarted a failed sample and chosed to remove the old files. Lets say I restarted sample 10, it then would remove all samples with 10 in the sample name.

I also found out that the sub handling the "which files to copy/delete" does so without knowing which sample to work on, it only has project-level information, so I was wondering if that could have an impact

[mariogiov]

The NGIProject object that is passed to the function as an argument has as children NGISample objects, and the removal function just parses through these. But the trouble is here:
https://github.com/NationalGenomicsInfrastructure/ngi_pipeline/blob/devel/ngi_pipeline/engines/piper_ngi/utils.py#L171
It's a naive glob that I guess assumed more unique sample names. This should be an easy fix though.

[Galithil]

yep, just did that. Now to hope that this still works...

On Mon, Jun 15, 2015 at 10:34 AM, Mario Giovacchini <
[email protected]> wrote:

The NGIProject object that is passed to the function as an argument has as
children NGISample objects, and the removal function just parses through
these. But the trouble is here:

https://github.com/NationalGenomicsInfrastructure/ngi_pipeline/blob/devel/ngi_pipeline/engines/piper_ngi/utils.py#L171
It's a naive glob that I guess assumed more unique sample names. This
should be an easy fix though.

—
Reply to this email directly or view it on GitHub
#298 (comment)
.