remux_plus

[dpark01]

Thinking of adding a new WDL workflow called remux_plus.

Inputs:

• old samplesheet (originally used)
• new samplesheet (as it ought to be)
• Array[File] raw_bams
• Array[File] cleaned_bams

Maybe the name is a bit of a misnomer, the goal would not be to actually call demux again, but instead to start with one task that takes the two samplesheets and turns them into a 3-col tab file for use with read_utils.reheader_bams. This file maps old sample names, library names, and filenames to new ones. It would then scatter invocations of reheader_bams on each raw_bam and cleaned_bam. Then would maybe re-run fastqc and multiqc and spike-in counts on all of them. Probably skip the other stuff (like spades and kraken.. though we could optionally).

The goal would be to reproduce most outputs of demux_plus without actually re-running it, when all that is desired is renaming samples and libraries and files based on a new samplesheet. This would obviously not suffice for actual changes to barcodes or read structures.

[tomkinsc]

Maybe it could consider the delta and actually re-demux any samples that do have barcode differences (perhaps discarding unmatched reads to save resources), in addition to re-headering any samples where the barcodes remain the same.