AlphaFold3 with ligands PAE json file: number of structure residues does not match matrix size

[rob.klose@…]

To whom it may concern:

When I load new alphafold3 predictions into chimera x and try to load the json files to look at PAE data I get the following error:

Structure fold_2024_05_09_15_13_cfp1_with_h3k4me3nuclesome_model_0.cif #1 does not match PAE matrix size 2020.The structure has 1998 polymer residues and 0 non-polymer atoms

This can happen if chains or atoms were deleted from the AlphaFold model or if the PAE data was applied to a structure that was not the one predicted by AlphaFold. Use the full-length AlphaFold model to show predicted aligned error.

It seems to only occur when I have post translational modifications in the prediction. Is there a fix for this?

Best wishes

Rob


Rob Klose

Professor of Genetics
Department of Biochemistry and Keble College
University of Oxford

www.kloselab.co.uk<http://www.kloselab.co.uk/>

PA:
Sarah-jane Scard
Sarah-jane.scard@bioch.ox.ac.uk<mailto:Sarah-jane.scard@bioch.ox.ac.uk>
Tel: 01865 613212

[Tom Goddard]

Yes it is fixed in the current ChimeraX daily build. AlphaFold 3 has PAE values for individual post-translational modification atoms while AlphaFold 2 only had PAE values for protein residues, so I had to update the ChimeraX code yesterday to make it work with AlphaFold 3 PAE.

[rob.klose@…]

Dear Tom

Thanks so much for your help with this.

I just uninstalled previous versions nad updated to version 1.9.dev202405150316 (2024-05-15). Unfortunately, I still get the same error message with structure predictions where I include PTMs.

The error message is this:

Structure fold_2024_05_10_13_38_hwdr82_ser5p_model_0.cif #1 does not match PAE matrix size 343.The structure has 334 polymer residues and 0 non-polymer atoms

This can happen if chains or atoms were deleted from the AlphaFold model or if the PAE data was applied to a structure that was not the one predicted by AlphaFold. Use the full-length AlphaFold model to show predicted aligned error.

Thanks again for all of your help.

Cheers

Rob



Rob Klose

Professor of Genetics
Department of Biochemistry and Keble College
University of Oxford

www.kloselab.co.uk

PA:
Sarah-jane Scard
Sarah-jane.scard@bioch.ox.ac.uk
Tel: 01865 613212





-----Original Message-----
From: ChimeraX <ChimeraX-bugs-admin@cgl.ucsf.edu> 
Sent: 15 May 2024 19:02
To: goddard@cgl.ucsf.edu; Rob Klose <rob.klose@bioch.ox.ac.uk>
Subject: Re: [ChimeraX] #15214: AlphaFold3 with ligands PAE json file: number of structure residues does not match matrix size

#15214: AlphaFold3 with ligands PAE json file: number of structure residues does not match matrix size
-------------------------------------------+-------------------------
          Reporter:  rob.klose@\u2026           |      Owner:  Tom Goddard
              Type:  defect                |     Status:  closed
          Priority:  normal                |  Milestone:
         Component:  Structure Prediction  |    Version:
        Resolution:  duplicate             |   Keywords:
        Blocked By:                        |   Blocking:
Notify when closed:                        |   Platform:  all
           Project:  ChimeraX              |
-------------------------------------------+-------------------------
Changes (by Tom Goddard):

 * resolution:   => duplicate
 * status:  assigned => closed

Comment:

 Yes it is fixed in the current ChimeraX daily build.  AlphaFold 3 has PAE  values for individual post-translational modification atoms while  AlphaFold 2 only had PAE values for protein residues, so I had to update  the ChimeraX code yesterday to make it work with AlphaFold 3 PAE.
--
Ticket URL: <https://www.rbvi.ucsf.edu/trac/ChimeraX/ticket/15214#comment:2>
ChimeraX <https://www.rbvi.ucsf.edu/chimerax/>
ChimeraX Issue Tracker

[Tom Goddard]

I think the one case I did not test was post-translational modification. Let me try that. Or if you want to send the .cif and .json I can investigate what is going on in your exact case.

[Tom Goddard]

Fixed in May 16 daily build or 1.8 release candidate.

AlphaFold 3 modified residues were not handled correctly by the ChimeraX PAE code. They are polymer residues but with atom-level PAE values. Made the pae plot and alphafold contacts command use atom-level PAE values for those.

[rob.klose@…]

Hi Tom,

It works perfectly now. Thanks for all of your help.

Cheers

Rob


-----Original Message-----
From: ChimeraX <ChimeraX-bugs-admin@cgl.ucsf.edu> 
Sent: Thursday, May 16, 2024 6:09 PM
To: goddard@cgl.ucsf.edu; Rob Klose <rob.klose@bioch.ox.ac.uk>
Subject: Re: [ChimeraX] #15214: AlphaFold3 with ligands PAE json file: number of structure residues does not match matrix size

#15214: AlphaFold3 with ligands PAE json file: number of structure residues does not match matrix size
-------------------------------------------+-------------------------
          Reporter:  rob.klose@\u2026           |      Owner:  Tom Goddard
              Type:  defect                |     Status:  closed
          Priority:  normal                |  Milestone:
         Component:  Structure Prediction  |    Version:
        Resolution:  duplicate             |   Keywords:
        Blocked By:                        |   Blocking:
Notify when closed:                        |   Platform:  all
           Project:  ChimeraX              |
-------------------------------------------+-------------------------
Comment (by Tom Goddard):

 I think the one case I did not test was post-translational modification.
 Let me try that.  Or if you want to send the .cif and .json I can  investigate what is going on in your exact case.
--
Ticket URL: <https://www.rbvi.ucsf.edu/trac/ChimeraX/ticket/15214#comment:4>
ChimeraX <https://www.rbvi.ucsf.edu/chimerax/>
ChimeraX Issue Tracker