Feature/support protein predictions

[KateSakharova]

Added support of using already predicted proteins as pipeline input.

• added script to split proteins.faa into HC, LC and PP files. Phophages processing is quite tricky because they have particular coords, not whole contig.
• fixed gff generator not to change coords for already predicted proteins for prophages (because they were initially predicted in contig coords)

[Ales-ibt]

It looks like it is hardcoded to use prodigal output as input with the option --use_proteins. I suppose it is OK for MGnify internal purposes, but I think it is a good idea to specify the expected format of such input for external users.

[Ales-ibt]

This was kind of difficult to read for me. You could evaluate whether the elements of two lists representing the ranges for prophage and protein intersect totally, partially, or not at all. The two lists are created using range and the intersection is checked:

def check_intersection(start_prophage, finish_prophage, start_protein, finish_protein):
    prophage_set = set(range(start_prophage, finish_prophage + 1))
    protein_set = set(range(start_protein, finish_protein + 1))

    intersection = prophage_set.intersection(protein_set)

    # Check for intersection size
    if len(intersection) > 0:
        if len(intersection) == len(protein_set):
            return "Protein inside prophage" 
        else:
            return "Protein intersects with prophage"
    else:
        return "Protein not in prophage"

[Ales-ibt]

In the MAP we use a gene coverage threshold of 75% as a criterion to tag the genes as part of the MGE or not. I think it is a good idea to homologate the criteria in both pipelines. What do you think?

[mberacochea]

We need to remember to update this dev -> master

[KateSakharova]

Yes, absolutely (bit now dev if quite different from master)

[Ales-ibt]

Thank you for this. I will update the protein coverage length on the next version of the MAP (wip) to be on the same page.

[Ales-ibt]

Protein length will always be shorter that the prophage length. Evaluating the protein coverage should be enough. In the MAP we evaluate the MGE length because we can have small MGEs as short as 1 Kb, but in the case of prophages, that shouldn't be the case.