Closed tomasbruna closed 1 year ago
Ahh, the previous implementation would not work if multiple separate protein files were on input. That's fixed.
The current implementation might still run into issues if the uptodate( [ $prot_seq_files[$i] ], [$prot_hintsfile]
is triggered. I'll look into this more.
Also, I realized IDs might clash if miniprot is run separately on multiple files (that's related to the original implementation as well). https://github.com/tomasbruna/miniprothint/commit/a38f3002f60bb47953988630bbeb26f04134ae9c should take care of that.
The uptodate stuff does not work in all places, anyway. I should go through the code and remove it... in the early days for BRAKER, when we had a lot of crashes, we had that idea that users would continue a failed run after fixing BRAKER... but as I said: it's buggy, either way.
The current implementation might still run into issues if the
uptodate( [ $prot_seq_files[$i] ], [$prot_hintsfile]
is triggered. I'll look into this more.
Our HPC is going into maintenance, tomorrow morning. I will run the tests on our end before merging. So the merge might take 2-3 days from now (depending on when our HPC awakens from its upgrade).
Thank you so much for all the work that you put into this, Tomas!!! This is really good.
The uptodate stuff does not work in all places, anyway. I should go through the code and remove it...
I was hoping you'd say that :), it seems like a big headache to maintain it.
If we at some point in time fully port to python, then snakemake (or Nextflow) would take care of this... timeline for this is very unsure on our end, though. I remove uptodate from the code for now.
I was hoping you'd say that :), it seems like a big headache to maintain it.
Tested on test1.sh.
Also tested on large protein inputs - "miniprothint gtf" was always matching the one miniprot would produce natively.