Hi Julian, AlphaFold 3 does assign different pLDDT values to each atom of standard residues. This is different from AlphaFold 2 where the exact same pLDDT values was assigned to every atom of a residue. I don't know what web site you are using to look at the pLDDT coloring and I don't know how that web site combines the pLDDT values for the atoms of a residue. You are correct that ChimeraX averages the pLDDT values of all atoms in the residue (backbone and side chain) when coloring ribbons by bfactor. As we discussed on this mailing list at length the AlphaFold coloring in discrete bands 50-70, 70-90 and 90-100 can create erratic and misleading coloring. if you have pLDDT values near 70 then you either get yellow, described as low confidence for slightly less than 70, or light blue described as high confidence for slightly more than 70. Kind of absurd to go from low to high confidence when values cross pLDDT 70. So ChimeraX instead smoothly interpolates colors for the pLDDT colors if you use "color bfactor palette alphafold". Unfortunately some researchers prefer to use the misleading discrete coloring. Tom
On Nov 8, 2024, at 7:22 AM, julian--- via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Here's the command with AF color hex codes from the AF3 server website
color bfactor palette 0,#ef821e:49.99,#ef821e:50,#f6ed12:69.99,#f6ed12:70,#10cff1:89.99,#10cff1:90,#106dff:100,#106dff key true
Now it's very close except I'm noticing that some small regions colored very high confidence by AF3 look just confident in Chimerax. Digging into it a bit, it seems the difference is that AF3 colors a residue by either the max bfactor of its atoms or maybe averages just the main-chain atoms. ChimeraX averages all atoms of a residue to color the cartoon, so a couple of C-delta atoms in the 80's is sometimes enough to bring the whole residue below 90. I'm guessing there's not way to match that AF behavior _______________________________________________ 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/