rnaseq icon indicating copy to clipboard operation
rnaseq copied to clipboard

Output CRAM as default

Open ewels opened this issue 3 years ago • 3 comments

Description of feature

We should probably be outputting CRAM files as default instead of BAM (it is an nf-core recommended feature in the guidelines, after all).

This was mentioned by @apeltzer to @drpatelh back in 2020 here and looks like it shouldn't be a huge amount of work?

Quote from above:

We could just update the main.nf on nf-core/modules to have optional outputs and amend the script section accordingly if required. Or the additional arguments to produce CRAM could also come via the options.args parameter provided by the developer.

CRAM outputs should save disk space for the end user and present few downsides. We can have a parameter to optionally still produce BAM instead if needed.

ewels avatar Jan 31 '22 13:01 ewels

Agree, it would be nice to output CRAM by default 👍🏽 Definitely not impossible but definitely not trivial either 😅

If we want to do it properly, we would have to test and update all relevant nf-core/local modules as well as perform conversions back to BAM for those tools that don't support it. May need to put aside a few days/week to sort it.

drpatelh avatar Feb 02 '22 16:02 drpatelh

Yup, agreed. Hopefully once a few pipelines start doing this it'll gain momentum because of the shared modules..

Sarek already does CRAM by default, right @maxulysse? So if we get the two most popular pipelines doing it, that should cover a lot of the most frequently used modules.

ewels avatar Feb 08 '22 23:02 ewels

I'd say we go for CRAM everywhere.

maxulysse avatar Feb 09 '22 06:02 maxulysse