Reopening PR

[DimaMolod]

New features/changes:

1. new flags in run_multimer_jobs._refactor.py and run_structure_prediction.py –dry_run, –skip_templates
2. run_structure_prediction.py: predict returns prediction_results
3. alphafold_backend.py: all postprocessing steps moved from prediction to dedicated postprocessing step
   1. Setup: added –allow_resume and –skip_templates
   2. Predict:
      1. Resume, read prediction_results from pickles and models from pdbs
      2. Removes templates if –skip_templates
      3. Adds unrelaxed proteins to prediction_results
      4. Stops after saving unrelaxed.pdb and result.pkl
   3. Postprocessing:
      1. Ranking
      2. Saves pae.json and plddt.json (<1 MB)
      3. Relaxation (incl. resume)
      4. AlphaFold style PAE plots
      5. Reports RMSD for TrueMultimer
      6. Zip pickles (add modelCIF and Konstantin’s truncator?)
4. minor changes:
   1. various logging levels including debug (clean logs)
   2. Always report AP version
   3. Report if the job was successful

[dingquanyu]

@DimaMolod could you resolve the conflicts in alphafold_backend.py

[dingquanyu]

@DimaMolod instead of changing the function back to the old plotting function here: https://github.com/KosinskiLab/AlphaPulldown/blob/eddb8648e7d655034b5d5571c25d4e6604f09406/alphapulldown/folding_backend/alphafold_backend.py#L461 I think you should keep using the new function plot_pae_from_matrix but inside plot_pae_from_matrix, you add options of using 2 colour schemes. old plotting function is not gonna be kept as it reads from results pickles.

[DimaMolod]

@DimaMolod instead of changing the function back to the old plotting function here:

https://github.com/KosinskiLab/AlphaPulldown/blob/eddb8648e7d655034b5d5571c25d4e6604f09406/alphapulldown/folding_backend/alphafold_backend.py#L461

I think you should keep using the new function plot_pae_from_matrix but inside plot_pae_from_matrix, you add options of using 2 colour schemes. old plotting function is not gonna be kept as it reads from results pickles.

it's not only the color scheme, but also the file names. Why do we need to name the plots after multimeric_object.description if the files are already in the directory named after that? I suggest we name it as e.g. PAE_plot_ranked_0_model_4_multimer_v3_pred_0.png I think it will be easier if you merge these as you see appropriate and simply commit