Adjust AlphaFold pLDDT coloring to better match EBI color scale.

[Tom Goddard]

From the ChimeraX mailing list:

​https://mail.cgl.ucsf.edu/mailman/archives/list/chimerax-users@cgl.ucsf.edu/thread/NUGJGHGBG7ZNY3KGULOQWXLFXYRJ3VI6/

Hi Vladimir,

Looking closer at the ChimeraX pLDDT colors I see they are not quite what I thought. Here are the ChimeraX colors

​https://www.cgl.ucsf.edu/chimerax/docs/user/commands/alphafold.html#colorConfidence

They are 100 dark blue, 90 light blue, 70 yellow, 50 orange, 0 red, and are linearly interpolated as I mentioned. The EBI discrete color ranges are

​https://www.ebi.ac.uk/training/online/courses/alphafold/inputs-and-outputs/evaluating-alphafolds-predicted-structures-using-confidence-scores/plddt-understanding-local-confidence/

Dark blue 90-100, light blue 70-90, yellow 50-70, and orange 0-50.

So you see ChimeraX matches the colors from the high end of the EBI color ranges, not from the middle of the EBI ranges as I suggested in my previous message. So the ChimeraX coloring is going to appear skewed in general to look less confident than the EBI coloring. I think the ChimeraX coloring would give a better match if it used 95 dark blue, 80 light blue, 60 yellow, 25 orange, 0 red and linearly interpolated. In other words use the middle value of the EBI ranges combined with the EBI colors. I've made a ChimeraX bug suggesting this change.

Tom

[Tom Goddard]

One complexity of this is that the ChimeraX and EBI colors are actually not the same RGB colors so some testing would be important to see if the overall impression is the same with the proposed adjustment of pLDDT values for each color.

[Tom Goddard]

The colors used in the EBI coloring differ from ChimeraX significantly. For instance EBI uses orange for 70 and has no yellow while ChimeraX uses yellow for 70. EBI uses orange-red for 0-50, and does not use red, while ChimeraX uses red for 0 and orange for 50. The blues are also different with EBI 70-90 being an aqua-blue while ChimeraX 90 (cornflower blue) is a lighter blue.

So I don't think these color schemes can be made "similar". I think it is best to stick with the current ChimeraX scheme since it has been use for 4 years now.

I added ChimeraX documentation (docs/user/alphafold-colors.html) describing both color schemes highlighting their differences. Elaine will link to this description from the relevant parts of the User Guide.

[Tom Goddard]

Eric suggests it is still worth adding a named ChimeraX palette that matches the EBI AlphaFold Database color scheme.

I tried to find the colors EBI uses but failed to find a definitive source when I was making the ChimeraX documentation page on our pLDDT coloring

​https://www.cgl.ucsf.edu/chimerax/docs/user/alphafold-colors.html

In the documentation I used colors I obtained from the Mac digital color meter app using the EBI plddt coloring page color key image

​https://www.ebi.ac.uk/training/online/courses/alphafold/inputs-and-outputs/evaluating-alphafolds-predicted-structures-using-confidence-scores/plddt-understanding-local-confidence/

I wasn't able to find a definitive source for the color RGB values in a quick search. If we had the exact RGB colors used by the EBI AlphaFold Database in its Mol* ribbon displays it would be pretty easy to add this extra palette. Although I discourage using it, I think it would do no harm to add such a palette since almost no users would find it and make use of it.