<!subteam^S016TBDBXHC> Dev meeting has started. I’m on workadventure. https://play.workadventu.re/@/nf-core/nf-core/office

hey, managed to implement it, and it runs as expected for a run of test_microbial --deduplication_mergejobs (this is the new param for running all contigs in one vs one-by-one) I need to do two additional things: 1. add to one of the test configs this param so that it gets checked 2. double check that the md5sum for merged/nonmerged deduplication is consistent (this should be the case)

nice!

i will open a PR likely next week after doing these last adjustments

for 2. it might not be the case that the bam is stable, because of the bam headers, which can include file paths for past commands btw. I worry that adding it to test_full might make the individual job too heavy for the GH runner, but we could try it, since it’s the one PR that’s still open. For actual benchmarking purposes I think we’d probably need to do two runs, one with and one without sharding/splitting, to compare to eager 2.* anyway

re 2) indeed -- the bam header changes things -- and strangely it seems the order of the flags in the reads themselves (so by-passing just by checking the .sam without the header doesn't work)

@Ian Light turns out test_full is already too heavy for the runners, and will need to be ran manually for releases only. That means you should add the new param to a different test.

will update this on this coming friday