Custom template fix, IPSAE, and Protenix #47
Conversation
…a_to_json. I need to update output_json logic to be a flag instead of a pathway
… set ipsae csv files to output for each model and a more descripting tool tip in the output page
…that it can be called from the cmd line
…om_template to add_custom_template
…path to af3_sif_path and added the option to the config ini, this permits other programs to be included in the future
… on ptm handling generally
| ipsae_scores=ipsae_scores, | ||
| verbose=False, | ||
| output_csv=ipsae_out) | ||
| except ValueError: |
There was a problem hiding this comment.
I'm not a big fan of blanket try, except blocks. It seems to be well-contained here but is there a reason for the value error that a if statement could catch?
There was a problem hiding this comment.
The ipsae code gave an error when running the ptm example with protenix, I then tested the ptm example with the other software and encountered some other bug, and so decided to just patch this for now with a try/except and have a proper look at what’s happening with ptms in the future.
There was a problem hiding this comment.
Could we instead add a check to see if there are PTMs present then not run the score?
There was a problem hiding this comment.
A PTM check is doable, but a little complicated to implement and likely to be removed in the next PR once I put some proper fixes in place.
I'm comfortable leaving this try except for now, but making an issue to fix it.
There was a problem hiding this comment.
Is there a smaller part of the code the try and except block can go over instead of the whole score? Could you also put in a logging error statement to explain what bugs are to be expected and to raise an issue on the git to report it and that the IPSAE scores will be disabled. Also, should we add the reference to the IPSAE score in the output page?
There was a problem hiding this comment.
Good idea to add in the ipsae reference! I've double checked and this is about as minimal I can make the try/except, but I have now added an error message.
|
Overall this is great! Very readable code good testing to accompany it. One thing I didn't see was the affinity prediction mode for Boltz2. This is a self-contained update however so it doesn't need to be included in this update. Once all comments have been addressed I'll approve the merge. |
abcfold/plots/pae-viewer-main/src/templates/paeViewerDemo__tab__citation.tpl
Outdated
Show resolved
Hide resolved
abcfold/plots/pae-viewer-main/src/templates/paeViewerDemo__tab__citation.tpl
Show resolved
Hide resolved
…reference and also fixed a small bug in ipsae that was causing ligand chain data to be missed
Key changes:
abcfold.pythey are now 2 different steps and this fixes the issue we were having.ipsae.pyto scripts - (and made it available at the cmd line in the toml). This is functionally the same as the ipsae github script, but I've made it into a class. I've also added support for several different types of PAE files tofile_handlers.pyso it can be used with af2, colabfold etc... I've also added this score into the output pages.backend_envs.py. Boltz and Chai are now installed in their own micromamba environments, but this retains the functionality we had before. We check to see if the env exists and the correct program version is installed, and install it if it isn't. I've also change how programs are called to accommodate this and made update the docs to reflect the change.