Also I have been using seqera co-scientist to investigate a parabricks issue and it pointed me towards this:

PARABRICKS_APPLYBQSR

gets

aln_idx

and

PARABRICKS_FQ2BAM.out.bqsr_table

as two separate queue channels without a

join

. They're matched by arrival order, not by

meta

. With several samples or lanes, a CRAM can end up paired with another sample's recalibration table. Joining them on meta first would fix that:

Copy code

aln_idx.join(PARABRICKS_FQ2BAM.out.bqsr_table)

I am now wondering if I needs to put everything for applybqsr into one input tuple. At some point I have heard that then forgot it again and now I am unsure haha.

do you have a link to the code line.

I think it refers to github.com/nf-core/sarek/blob/…/main.nf#…

ah yes: github.com/nf-core/sarek/blob/…/main.nf#… needcs to have the same pattern as here

I will fix it 🙂