Interaction screening - one-to-all, pairwise multimer predictions

[keiran-rowell-unsw]

Description of feature

Handle the samplesheet such that the user can provide either a) a reference protein and then set of other proteins to co-fold with, or b) a collection of individual proteins to co-fold all pairs looking for interactions.

Hopefully the channel flow can be repurposed using .map statements with a Nextflow filter, a la:

fasta_files
            .filter { sample != reference }
            .map { sample -> tuple("${reference.name}_vs_${sample.name}", [reference, sample]) }

And

 fasta_files
            .combine(fasta_files)
            .filter { file1 != file2 }
            .map { file1, file2 -> tuple("${file1.name}_vs_${file2.name}", [file1, file2]) }

It would be nice to then extract the residue pair interaction data (PAE) and interface interaction score (iPTM) to 'rank' which proteins might be interacting.

These could be extracted into a MultiQC 'summary' report (not the nice visualisations of each protein structure, but purely a summary file to pass to MultiQC). Might want extensive testing before pushing back as a main MultiQC module

Australian-Structural-Biology-Computing/MultiQC@4b8fa20