Make Prodigal and Diamond outputs reusable.

[JeanMainguy]

Make Prodigal and Diamond outputs reusable.

[chklovski]

I'll look into implementing a '--resume' function after diamond has been run - this might also be useful if users want to query the results of a different completeness model when they didn't initially specify it. I probably won't get to this for ~ a month though. In the mean time, if you are running checkM2 on identical bins, you can save the results of the output and reuse it? If the contigs in the bins change, the checkm2 results will also change.

[JeanMainguy]

In my case contigs do not change (their sequences and ids do not change) but their repartitions in bins change making checkM2 prediction change as well. As the contigs do not change, it would be faster to launch Diamond only once and use its result to analyze all bin sets.

I could try to make it work with the resume flag. For that I would need to launch prodigal and diamond on all contigs and shape their results specifically for each bin set mimicking what checkM2 would produced. Then I launch checkM2 with the resume flag and hopefully it will resume. It would not be the cleanest option but it would work.

[JeanMainguy]

I can try to implement a first version of the --resume flag and make a pull request if you are alright with that?