How to show AlphaFold error estimates with ChimeraX
ฝัง
- เผยแพร่เมื่อ 11 ก.ย. 2024
- We take a look at how to show AlphaFold predicted aligned error (PAE) plots to see if domains are packed correctly, and also show how to color by AlphaFold per-residue confidence scores. The PAE plots are in ChimeraX versions from April 2022 (not in version 1.3). More details here
www.rbvi.ucsf....
I run command "color bfactor #1 palette alphafold", but I think the color is not so good. How can I modify or redefine the palette alphafold?
Hey thanks so much for the help, quick question: what is the different between color PAE domain and color pLDDT ?
Which pallete coloring format is the "Color PAE domains" using? esmfold, alphafold, paecontacts, rainbow?
How do you open PAE data from a model created by AlphaColab on Chimera? After I upload the 'predicted_aligned_error_v1.json" JSON file', I get an error saying JSON file 'X' is not AlphaFold predicted aligned error data, expected a top level list PAE file suffix must be ".json" or ".pkl".
if you are multimer the file should be "scores_rank_001_alphafold2_multimer_v3_model_4_seed_000.json"
Is this available in Chimera X 1.3 for mac? I don't have an option for pae plots in tools/Structure Prediction/Alphafold
The error plot was added April 2022 and is in the ChimeraX daily builds. It is not in ChimeraX 1.3 which is from November 2021.
Hi! I have tried to calculate the model of a protein complex several times now in order to explore this feature, but I am always getting the same error message - "Minimization failed after 100 attempts". I have also tried to calculate models that I was able to calculate before and they too show the same error message. I have tried to install the most recent daily build of Chimera X, but the problem remains. Can you help me? Thank you!
AlphaFold uses OpenMM for energy minimization with the Amber force field. If the predicted structure has bad clashes minimization will often fail as you have observed. In the ChimeraX daily build the AlphaFold panel has an Options button "Energy minimize predicted structures". Turn that off and rerun your prediction. I thought I made ChimeraX give you the unminimized structure if minimization fails but I'll have to check. If I haven't done that I'll try to update the code so at least you get the unminimized structure. With bad clashes there is no simple way to minimize. You might be able to fix the problems by hand moving residues and then minimizing.
I looked at the ChimeraX AlphaFold Google Colab code and it does not handle OpenMM minimization failures. So I've made a request to improve it so it provides the unminimized structure instead of just failing. www.rbvi.ucsf.edu/trac/ChimeraX/ticket/6945
@@ucsfchimerax8387 Thank you so much for the fast and very complete answer. The version of Chimera X I currently own on my computer has no "Options" button on the AlphaFold panel, but I will install the most recent daily build. The request you made is fantastic and if I could make a suggestion, I think it would be great if Chimera X could provide the 5 unminimized structures, not only when the minimization step fails, but also when the Google Colab server runtime is terminated in that particular step. It is frustrating knowing you were so close of getting the final product, to just end up with nothing when the unminimized models are right there.
Nevertheless, and in order to confirm that this error is dependent on the predicted structure, I tried to predict the structure of a 70 amino acid protein whose model has been calculated several times through AlphaFold in classes by many students, and the error still pops up:
ValueError Traceback (most recent call last)
in ()
692 seq_list = seq_list[1:]
693
--> 694 run_prediction(seq_list, energy_minimize = not dont_minimize)
5 frames
/usr/local/lib/python3.7/dist-packages/alphafold/relax/amber_minimize.py in _run_one_iteration(pdb_string, max_iterations, tolerance, stiffness, restraint_set, max_attempts, use_gpu, exclude_residues)
417 logging.info(e)
418 if not minimized:
--> 419 raise ValueError(f"Minimization failed after {max_attempts} attempts.")
420 ret["opt_time"] = time.time() - start
421 ret["min_attempts"] = attempts
ValueError: Minimization failed after 100 attempts.
@@tomasfernandes7944 Yes a fixed version would give you all 5 unminimized structures. The ChimeraX AlphaFold Colab script only tries to minimize the best scoring structure of the 5 and by the time it tries that minimization it already has 5 unminimized structures computed.