KosinskiLab / AlphaPulldown

https://doi.org/10.1093/bioinformatics/btac749
GNU General Public License v3.0
199 stars 46 forks source link

Pad input features from different prediction jobs to a universal desired shapes #299

Closed dingquanyu closed 6 months ago

dingquanyu commented 6 months ago

1) pad input matrices based on desired number of residues and desired number of msas 2) now can run multiple jobs on one GPU

TODO: 1) check model_name_flags 2) fix the postprocessing step when template sequence and query sequence are not the same in multimeric mode

dingquanyu commented 6 months ago

model_names flag be a DEFINE_list instead of DEFINE_string

dingquanyu commented 6 months ago

@maurerv could you double check the run_structure_prediction.py and alphafold_backend.py specifically? Besides, I've removed unifold_backend as we are not gonna use them now that we have OpenFold. So the change in the predict() within run_structure_prediction.py should not be a problem.

maurerv commented 6 months ago

The changes to run_structure_prediction and alphafold_backend do not conflict with the pipeline. Feel free to proceed :)

dingquanyu commented 6 months ago

The changes to run_structure_prediction and alphafold_backend do not conflict with the pipeline. Feel free to proceed :)

Thanks for the review!