[ali-yz]

Adds an optional protein trace (--protein_trace true) that records, for every input protein, what happened to it during the run, and writes:

• protein_fates.tsv: one row per input protein with exactly one final fate (e.g. in_orthoxml, too_short, lost_in_split_hog, singleton_roothog_dropped, ...)
• protein_events.tsv.gz: all recorded events, sorted by species and protein
• protein_trace_summary.txt: counts per fate, the rootHOGs that lose the most proteins, and consistency checks

How it works

• infer_roothogs and infer_subhogs write per-task protein_events_*.events files, but only when the new --protein-trace option is given. Nextflow passes it when params.protein_trace is set; it is off by default, and then nothing is written and no extra work is done.
• A new protein_trace process merges the event files at the end. It sorts them on disk (GNU sort) and handles one protein at a time, so memory stays at a few hundred bytes per protein. Its memory request scales with the size of the event files (conf/base.config).

How to use it

To turn it on, add --protein_trace true to your Nextflow command. e.g.

nextflow run FastOMA.nf -profile docker \
    --container_version "sha-$(git rev-list --max-count=1 --abbrev-commit HEAD)" \
    --input testdata/in_folder \
    --output_folder myresult/
    --protein_trace true

[alpae]

Hi @ali-yz ,

PR looks already quite good. I found 3 points that should be fixed, (two easy, one might need a bit more thinking).

I hope it's clear what I mean, otherwise ping me, np!

[alpae]

the broken tests are unrelated. I fixed them in #107. If you rebase, things should become green again ;-)

[ali-yz]

@alpae Thank you so much for your comments, I believe they are addressed now. You can go ahead with the merge whenever ready.