AlphaFold3 structures coloured in ChimeraX
Hello, I am using ChimeraX 10.1 to display the results of AlphaFold3 predictions. I observe a large discrepancy between the colour coding of the quality of the structure (pIDDT) displayed at the output of the AF3 server and ChimeraX (I use the command color bfactor #1 palette AlphaFold, 1 being thenumbero of the structure to colour). ChimeraX's colouring indicates that the quality of the predicted structure is substantially lower then than that of theAF3 server. Is there any possibility to get ChimeraX to display the AF3 server colours? Below is an example, right AF3 output, left ChimedraX. Best regards Vladimir
TRY THIS: ChimeraX Command line to match pLDDT of AF3: color bfactor palette 0,#FF7D45:50,#FF7D45:50.00001,#FFDB13:70,#FFDB13:70.00001,#65C BF3:90,#65CBF3:90.00001,#0053D6:100,#0053D6 Dr Khawar Sohail Siddiqui, [cid:77069661-1c4a-498b-b269-a936e3f74d9a]hoo.com ________________________________ From: Vladimir Saudek via ChimeraX-users <chimerax-users@cgl.ucsf.edu> Sent: Thursday, September 11, 2025 4:11 PM To: chimerax-users@cgl.ucsf.edu <chimerax-users@cgl.ucsf.edu> Subject: [chimerax-users] AlphaFold3 structures coloured in ChimeraX Hello, I am using ChimeraX 10.1 to display the results of AlphaFold3 predictions. I observe a large discrepancy between the colour coding of the quality of the structure (pIDDT) displayed at the output of the AF3 server and ChimeraX (I use the command color bfactor #1 palette AlphaFold, 1 being thenumbero of the structure to colour). ChimeraX's colouring indicates that the quality of the predicted structure is substantially lower then than that of theAF3 server. Is there any possibility to get ChimeraX to display the AF3 server colours? Below is an example, right AF3 output, left ChimedraX. Best regards Vladimir [cid:part1.3xWabto1.xnDhu9y4@gmail.com][cid:part2.OY11MAlU.YesutBDI@gmail.com] This email and any files transmitted with it may contain confidential information. If you believe you have received this email or any of its contents in error, please notify me immediately by return email and destroy this email. Do not use, disseminate, forward, print or copy any contents of an email received in error.
Yes, this is the way. Best, Jan
On 11. Sep 2025, at 08:21, Khawar Siddiqui via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
TRY THIS: ChimeraX Command line to match pLDDT of AF3: color bfactor palette 0,#FF7D45:50,#FF7D45:50.00001,#FFDB13:70,#FFDB13:70.00001,#65C BF3:90,#65CBF3:90.00001,#0053D6:100,#0053D6
Dr Khawar Sohail Siddiqui, <image.png>hoo.com <http://hoo.com/> From: Vladimir Saudek via ChimeraX-users <chimerax-users@cgl.ucsf.edu <mailto:chimerax-users@cgl.ucsf.edu>> Sent: Thursday, September 11, 2025 4:11 PM To: chimerax-users@cgl.ucsf.edu <mailto:chimerax-users@cgl.ucsf.edu> <chimerax-users@cgl.ucsf.edu <mailto:chimerax-users@cgl.ucsf.edu>> Subject: [chimerax-users] AlphaFold3 structures coloured in ChimeraX
Hello, I am using ChimeraX 10.1 to display the results of AlphaFold3 predictions. I observe a large discrepancy between the colour coding of the quality of the structure (pIDDT) displayed at the output of the AF3 server and ChimeraX (I use the command color bfactor #1 palette AlphaFold, 1 being thenumbero of the structure to colour). ChimeraX's colouring indicates that the quality of the predicted structure is substantially lower then than that of theAF3 server. Is there any possibility to get ChimeraX to display the AF3 server colours? Below is an example, right AF3 output, left ChimedraX. Best regards Vladimir
<PF1cTdy9lAAAM3U5.png><Qrb7tniUSn9p3ZNB.png>
This email and any files transmitted with it may contain confidential information. If you believe you have received this email or any of its contents in error, please notify me immediately by return email and destroy this email. Do not use, disseminate, forward, print or copy any contents of an email received in error.
_______________________________________________ ChimeraX-users mailing list -- chimerax-users@cgl.ucsf.edu <mailto:chimerax-users@cgl.ucsf.edu> To unsubscribe send an email to chimerax-users-leave@cgl.ucsf.edu <mailto:chimerax-users-leave@cgl.ucsf.edu> Archives: https://mail.cgl.ucsf.edu/mailman/archives/list/chimerax-users@cgl.ucsf.edu/
You can adjust the palette by tweaking the range, eg: color bfactor #1 palette alphafold range .1,1 ________________________________ From: Vladimir Saudek via ChimeraX-users <chimerax-users@cgl.ucsf.edu> Sent: 11 September 2025 18:11 To: chimerax-users@cgl.ucsf.edu <chimerax-users@cgl.ucsf.edu> Subject: [chimerax-users] AlphaFold3 structures coloured in ChimeraX Hello, I am using ChimeraX 10.1 to display the results of AlphaFold3 predictions. I observe a large discrepancy between the colour coding of the quality of the structure (pIDDT) displayed at the output of the AF3 server and ChimeraX (I use the command color bfactor #1 palette AlphaFold, 1 being thenumbero of the structure to colour). ChimeraX's colouring indicates that the quality of the predicted structure is substantially lower then than that of theAF3 server. Is there any possibility to get ChimeraX to display the AF3 server colours? Below is an example, right AF3 output, left ChimedraX. Best regards Vladimir [cid:part1.3xWabto1.xnDhu9y4@gmail.com][cid:part2.OY11MAlU.YesutBDI@gmail.com]
Hi Vladimir, This has been discussed multiple times on the ChimeraX mailing list, for example, here https://mail.cgl.ucsf.edu/mailman/archives/list/chimerax-users@cgl.ucsf.edu/... The question always arises in the same context -- the pLDDT colored AlphaFold model looks worse in ChimeraX then with the web server coloring. That is because your model is slightly above the discrete cutoffs for the pLDDT color scale used by the web server. If your residues are above 70 you get blue, but below 70 you get yellow. The discrete levels mislead you, either making the structure prediction look better because many residues are slighly above 70 or making it look worse because many are slightly below 70. ChimeraX instead uses a linearly interpolated scale where 70 uses a color half-way between blue and yellow. This gives a less misleading view of the confidence. But everyone who has raised this problem wants their prediction to look higher confidence so they want the discrete color cutoffs. No one has yet raised the issue where the discrete cutoffs made it look less confident than ChimeraX due to pLDDT in the high 60s. I suggest not using the misleading discrete scale (especially where it makes your prediction look more confident than the actual pLDDT values due to the specific cutoff of 70). In past discussions people feel the discrete colors are the standard promoted by EBI and the AlphaFold developers and don't seem to mind that it is sometimes misleading. Tom
On Sep 10, 2025, at 11:11 PM, Vladimir Saudek via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Hello, I am using ChimeraX 10.1 to display the results of AlphaFold3 predictions. I observe a large discrepancy between the colour coding of the quality of the structure (pIDDT) displayed at the output of the AF3 server and ChimeraX (I use the command color bfactor #1 palette AlphaFold, 1 being thenumbero of the structure to colour). ChimeraX's colouring indicates that the quality of the predicted structure is substantially lower then than that of theAF3 server. Is there any possibility to get ChimeraX to display the AF3 server colours? Below is an example, right AF3 output, left ChimedraX. Best regards Vladimir
<PF1cTdy9lAAAM3U5.png><Qrb7tniUSn9p3ZNB.png>_______________________________________________ ChimeraX-users mailing list -- chimerax-users@cgl.ucsf.edu <mailto:chimerax-users@cgl.ucsf.edu> To unsubscribe send an email to chimerax-users-leave@cgl.ucsf.edu <mailto:chimerax-users-leave@cgl.ucsf.edu> Archives: https://mail.cgl.ucsf.edu/mailman/archives/list/chimerax-users@cgl.ucsf.edu/
Hi Vladimir, Looking closer at the ChimeraX pLDDT colors I see they are not matching the EBI scale the way I thought. Here are the ChimeraX colors https://www.cgl.ucsf.edu/chimerax/docs/user/commands/alphafold.html#colorCon... They are 100 dark blue, 90 light blue, 70 yellow, 50 orange, 0 red, and are linearly interpolated. The EBI discrete color ranges are https://www.ebi.ac.uk/training/online/courses/alphafold/inputs-and-outputs/e... 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. https://www.rbvi.ucsf.edu/trac/ChimeraX/ticket/18769 Tom
On Sep 11, 2025, at 1:44 PM, Tom Goddard via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Hi Vladimir,
This has been discussed multiple times on the ChimeraX mailing list, for example, here
https://mail.cgl.ucsf.edu/mailman/archives/list/chimerax-users@cgl.ucsf.edu/...
The question always arises in the same context -- the pLDDT colored AlphaFold model looks worse in ChimeraX then with the web server coloring. That is because your model is slightly above the discrete cutoffs for the pLDDT color scale used by the web server. If your residues are above 70 you get blue, but below 70 you get yellow. The discrete levels mislead you, either making the structure prediction look better because many residues are slighly above 70 or making it look worse because many are slightly below 70. ChimeraX instead uses a linearly interpolated scale where 70 uses a color half-way between blue and yellow. This gives a less misleading view of the confidence. But everyone who has raised this problem wants their prediction to look higher confidence so they want the discrete color cutoffs. No one has yet raised the issue where the discrete cutoffs made it look less confident than ChimeraX due to pLDDT in the high 60s.
I suggest not using the misleading discrete scale (especially where it makes your prediction look more confident than the actual pLDDT values due to the specific cutoff of 70). In past discussions people feel the discrete colors are the standard promoted by EBI and the AlphaFold developers and don't seem to mind that it is sometimes misleading.
Tom
On Sep 10, 2025, at 11:11 PM, Vladimir Saudek via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Hello, I am using ChimeraX 10.1 to display the results of AlphaFold3 predictions. I observe a large discrepancy between the colour coding of the quality of the structure (pIDDT) displayed at the output of the AF3 server and ChimeraX (I use the command color bfactor #1 palette AlphaFold, 1 being thenumbero of the structure to colour). ChimeraX's colouring indicates that the quality of the predicted structure is substantially lower then than that of theAF3 server. Is there any possibility to get ChimeraX to display the AF3 server colours? Below is an example, right AF3 output, left ChimedraX. Best regards Vladimir
<PF1cTdy9lAAAM3U5.png><Qrb7tniUSn9p3ZNB.png>_______________________________________________ ChimeraX-users mailing list -- chimerax-users@cgl.ucsf.edu <mailto:chimerax-users@cgl.ucsf.edu> To unsubscribe send an email to chimerax-users-leave@cgl.ucsf.edu <mailto:chimerax-users-leave@cgl.ucsf.edu> Archives: https://mail.cgl.ucsf.edu/mailman/archives/list/chimerax-users@cgl.ucsf.edu/
_______________________________________________ ChimeraX-users mailing list -- chimerax-users@cgl.ucsf.edu To unsubscribe send an email to chimerax-users-leave@cgl.ucsf.edu Archives: https://mail.cgl.ucsf.edu/mailman/archives/list/chimerax-users@cgl.ucsf.edu/
Hi Tom, Many thanks for the clear explanation. Your linearly interpolated scale has lots of merits but it might be difficult to get it accepted as a standard. Perhaps if Chimera X could offer both options it would let the users see the obvious superiority of your approach. Best wishes Vladimir On 11/09/2025 22:44, Tom Goddard wrote:
Hi Vladimir,
This has been discussed multiple times on the ChimeraX mailing list, for example, here
https://mail.cgl.ucsf.edu/mailman/archives/list/chimerax-users@cgl.ucsf.edu/...
The question always arises in the same context -- the pLDDT colored AlphaFold model looks worse in ChimeraX then with the web server coloring. That is because your model is slightly above the discrete cutoffs for the pLDDT color scale used by the web server. If your residues are above 70 you get blue, but below 70 you get yellow. The discrete levels mislead you, either making the structure prediction look better because many residues are slighly above 70 or making it look worse because many are slightly below 70. ChimeraX instead uses a linearly interpolated scale where 70 uses a color half-way between blue and yellow. This gives a less misleading view of the confidence. But everyone who has raised this problem wants their prediction to look higher confidence so they want the discrete color cutoffs. No one has yet raised the issue where the discrete cutoffs made it look less confident than ChimeraX due to pLDDT in the high 60s.
I suggest not using the misleading discrete scale (especially where it makes your prediction look more confident than the actual pLDDT values due to the specific cutoff of 70). In past discussions people feel the discrete colors are the standard promoted by EBI and the AlphaFold developers and don't seem to mind that it is sometimes misleading.
Tom
On Sep 10, 2025, at 11:11 PM, Vladimir Saudek via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Hello, I am using ChimeraX 10.1 to display the results of AlphaFold3 predictions. I observe a large discrepancy between the colour coding of the quality of the structure (pIDDT) displayed at the output of the AF3 server and ChimeraX (I use the command color bfactor #1 palette AlphaFold, 1 being thenumbero of the structure to colour). ChimeraX's colouring indicates that the quality of the predicted structure is substantially lower then than that of theAF3 server. Is there any possibility to get ChimeraX to display the AF3 server colours? Below is an example, right AF3 output, left ChimedraX. Best regards Vladimir
<PF1cTdy9lAAAM3U5.png><Qrb7tniUSn9p3ZNB.png>_______________________________________________ ChimeraX-users mailing list --chimerax-users@cgl.ucsf.edu To unsubscribe send an email tochimerax-users-leave@cgl.ucsf.edu Archives:https://mail.cgl.ucsf.edu/mailman/archives/list/chimerax-users@cgl.ucsf.edu/
Just a quick thought: Wouldn't it be useful to scale the colouring acording to “atom bfactor range MIN to MAX instead of 0 to 1? I obtained these numbers from the Log after the command "color bfactor #1 palette alphafold". and then used them in: "color bfactor @ca palette alphafold range 0.186,0.91 target c key true". On 12/09/2025 00:40, Tom Goddard wrote:
Hi Vladimir,
Looking closer at the ChimeraX pLDDT colors I see they are not matching the EBI scale the way I thought. Here are the ChimeraX colors
https://www.cgl.ucsf.edu/chimerax/docs/user/commands/alphafold.html#colorCon...
They are 100 dark blue, 90 light blue, 70 yellow, 50 orange, 0 red, and are linearly interpolated. The EBI discrete color ranges are
https://www.ebi.ac.uk/training/online/courses/alphafold/inputs-and-outputs/e...
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.
https://www.rbvi.ucsf.edu/trac/ChimeraX/ticket/18769
Tom
On Sep 11, 2025, at 1:44 PM, Tom Goddard via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Hi Vladimir,
This has been discussed multiple times on the ChimeraX mailing list, for example, here
https://mail.cgl.ucsf.edu/mailman/archives/list/chimerax-users@cgl.ucsf.edu/...
The question always arises in the same context -- the pLDDT colored AlphaFold model looks worse in ChimeraX then with the web server coloring. That is because your model is slightly above the discrete cutoffs for the pLDDT color scale used by the web server. If your residues are above 70 you get blue, but below 70 you get yellow. The discrete levels mislead you, either making the structure prediction look better because many residues are slighly above 70 or making it look worse because many are slightly below 70. ChimeraX instead uses a linearly interpolated scale where 70 uses a color half-way between blue and yellow. This gives a less misleading view of the confidence. But everyone who has raised this problem wants their prediction to look higher confidence so they want the discrete color cutoffs. No one has yet raised the issue where the discrete cutoffs made it look less confident than ChimeraX due to pLDDT in the high 60s.
I suggest not using the misleading discrete scale (especially where it makes your prediction look more confident than the actual pLDDT values due to the specific cutoff of 70). In past discussions people feel the discrete colors are the standard promoted by EBI and the AlphaFold developers and don't seem to mind that it is sometimes misleading.
Tom
On Sep 10, 2025, at 11:11 PM, Vladimir Saudek via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Hello, I am using ChimeraX 10.1 to display the results of AlphaFold3 predictions. I observe a large discrepancy between the colour coding of the quality of the structure (pIDDT) displayed at the output of the AF3 server and ChimeraX (I use the command color bfactor #1 palette AlphaFold, 1 being thenumbero of the structure to colour). ChimeraX's colouring indicates that the quality of the predicted structure is substantially lower then than that of theAF3 server. Is there any possibility to get ChimeraX to display the AF3 server colours? Below is an example, right AF3 output, left ChimedraX. Best regards Vladimir
<PF1cTdy9lAAAM3U5.png><Qrb7tniUSn9p3ZNB.png>_______________________________________________ ChimeraX-users mailing list --chimerax-users@cgl.ucsf.edu To unsubscribe send an email tochimerax-users-leave@cgl.ucsf.edu Archives:https://mail.cgl.ucsf.edu/mailman/archives/list/chimerax-users@cgl.ucsf.edu/
_______________________________________________ ChimeraX-users mailing list -- chimerax-users@cgl.ucsf.edu To unsubscribe send an email to chimerax-users-leave@cgl.ucsf.edu Archives: https://mail.cgl.ucsf.edu/mailman/archives/list/chimerax-users@cgl.ucsf.edu/
Although using the actual min,max will get the most dynamic range of coloring, you would no longer get a consistent coloring of the same value between different structures. If the goal is to easily understand the confidence scores when you see the colors, and compare them between different structures, you need a consistent value-color mapping. Elaine ----- Elaine C. Meng, Ph.D. UCSF Chimera(X) team Resource for Biocomputing, Visualization, and Informatics Department of Pharmaceutical Chemistry University of California, San Francisco
On Sep 12, 2025, at 2:09 AM, Vladimir Saudek via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Just a quick thought: Wouldn't it be useful to scale the colouring acording to “atom bfactor range MIN to MAX instead of 0 to 1? I obtained these numbers from the Log after the command "color bfactor #1 palette alphafold". and then used them in: "color bfactor @ca palette alphafold range 0.186,0.91 target c key true".
On 12/09/2025 00:40, Tom Goddard wrote:
Hi Vladimir,
Looking closer at the ChimeraX pLDDT colors I see they are not matching the EBI scale the way I thought. Here are the ChimeraX colors
https://www.cgl.ucsf.edu/chimerax/docs/user/commands/alphafold.html#colorCon...
They are 100 dark blue, 90 light blue, 70 yellow, 50 orange, 0 red, and are linearly interpolated. The EBI discrete color ranges are
https://www.ebi.ac.uk/training/online/courses/alphafold/inputs-and-outputs/e...
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.
https://www.rbvi.ucsf.edu/trac/ChimeraX/ticket/18769
Tom
On Sep 11, 2025, at 1:44 PM, Tom Goddard via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Hi Vladimir,
This has been discussed multiple times on the ChimeraX mailing list, for example, here
https://mail.cgl.ucsf.edu/mailman/archives/list/chimerax-users@cgl.ucsf.edu/...
The question always arises in the same context -- the pLDDT colored AlphaFold model looks worse in ChimeraX then with the web server coloring. That is because your model is slightly above the discrete cutoffs for the pLDDT color scale used by the web server. If your residues are above 70 you get blue, but below 70 you get yellow. The discrete levels mislead you, either making the structure prediction look better because many residues are slighly above 70 or making it look worse because many are slightly below 70. ChimeraX instead uses a linearly interpolated scale where 70 uses a color half-way between blue and yellow. This gives a less misleading view of the confidence. But everyone who has raised this problem wants their prediction to look higher confidence so they want the discrete color cutoffs. No one has yet raised the issue where the discrete cutoffs made it look less confident than ChimeraX due to pLDDT in the high 60s.
I suggest not using the misleading discrete scale (especially where it makes your prediction look more confident than the actual pLDDT values due to the specific cutoff of 70). In past discussions people feel the discrete colors are the standard promoted by EBI and the AlphaFold developers and don't seem to mind that it is sometimes misleading.
Tom
On Sep 10, 2025, at 11:11 PM, Vladimir Saudek via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Hello, I am using ChimeraX 10.1 to display the results of AlphaFold3 predictions. I observe a large discrepancy between the colour coding of the quality of the structure (pIDDT) displayed at the output of the AF3 server and ChimeraX (I use the command color bfactor #1 palette AlphaFold, 1 being thenumbero of the structure to colour). ChimeraX's colouring indicates that the quality of the predicted structure is substantially lower then than that of theAF3 server. Is there any possibility to get ChimeraX to display the AF3 server colours? Below is an example, right AF3 output, left ChimedraX. Best regards Vladimir
participants (6)
-
Adam Middleton -
Elaine Meng -
Jan Kosinski -
Khawar Siddiqui -
Tom Goddard -
Vladimir Saudek