https://nf-co.re logo
Join Slack
Powered by
<@U01T3NSK7TJ> I improved the CI in <https://githu...
# createpanelrefs
m
@Ram I improved the CI in https://github.com/nf-core/createpanelrefs/pull/48 ✅ + merged (already approved and merged) In https://github.com/nf-core/createpanelrefs/pull/49 I added tests for trying to reproduce and solve our friend Francesco issue with mutect2 PON, but the issue is still under investigation. I simplified the usage of the references in tests, hope it's still understandable
49 was merged as well. I don't see anything weird in @Francesco Lescai’s config file, and the changes I made in 49 are truly replicating igenomes usage
@Ram should we have a meeting today? I don't remember if we said every 2 weeks or every months...
f
I’ll try to launch again and send the log before lunch, including just a test profile
m
2 questions for you @Francesco Lescai, do you specify a 
mutect2_target_bed
params?
f
I haven’t (did send the launch command didn’t I?)
m
have you tried the same command with no fai and no dict (
--fai null --dict null
)
f
also have not (see launch command) but can do to test it if you like
(have some meetings back to back will be available again later)
m
no worries, same thing here
Let me try out if it's necessary
ok, so it fails without 
mutect2_target_bed
ERROR ~ Invalid method invocation 
call
with arguments: [[id:test], [/home/max/tests/3ef06401100c247c87b266da0ba933cb/work/98/3bd293b80a6ffe1cab775e3283fc60/sample1.vcf.gz, /home/max/tests/3ef06401100c247c87b266da0ba933cb/work/1a/c44bab32dbf419d02835d164f40493/sample2.vcf.gz], [/home/max/tests/3ef06401100c247c87b266da0ba933cb/work/98/3bd293b80a6ffe1cab775e3283fc60/sample1.vcf.gz.tbi, /home/max/tests/3ef06401100c247c87b266da0ba933cb/work/1a/c44bab32dbf419d02835d164f40493/sample2.vcf.gz.tbi], /nf-core/test-datasets/modules/data/genomics/homo_sapiens/genome/chr21/sequence/genome.dict] (java.util.LinkedList) on _closure6 type
ok
no need to try with fai and dict empty
I think I know where this is coming from
f
it makes sense to enforce 
mutect2_target_bed
in this case it’s a panel, so I should have thought about it anyway. I was thinking about setting the PoN to include off-target stuff in order to allow them to be flagged. but I also suspect they’d fail several intervals for either lack of data or missing some filters along the way. it should not be directly related to this error, but it does point out to a scientifically important element of this workflow we might want to explicitly mention in the docs (I’m happy to write this part, if it helps)
m
Let me add an auto-generation of target_bed based on the fai, and from there we can improve
🤩 2
r
@Maxime Garcia I think we agreed to meet once every two weeks. However, the meeting invite does not seem to reflect this as a recurring event. Can you update the meeting invite? 😄
✅ 1
Also, I don't use mutect2 that often. But does it make sense to have checks in place to error and exit (or issue warnings) if we are missing a mandatory parameter? Something like this or with a better different approach? 😄
This could be done on top of updating the docs like Francesco suggested.
m
I was thinking into moving the references bits out of the subworkflows and into its own to avoid process duplication
And yes updating docs and error messages
We'll talk more during meeting
👍 1
r
Sure! Let's do that 😄
f
I’ll join as well. I’ve sent privately to maxime the logs. running the test profile doesn’t help because it’s set the tool to cnvkit and it seems to work so haven’t sent it.
m
f
still 14:30 yes? I have it in the calendar unless it’s an old schedule
m
oh, so #C05GH685FKJ are 13:30, #C01NHS4CC91 are #14:00, #CGFUX04HZ are 1430
BAD COMMUNICATION on my part today, definitively facepalm
f
ah jesus 😅 give me 5 minutes I just finished lunch with my lab, I’ll go back to my desk
Open in Slack
PreviousNext
Powered by