[chimerax-users] Analysis of the protein-ligand docking results with the X-ray data

[Enrico Martinez]

Hello Elaine!
Many thanks for these excellent ideas!
With kind regards,
Enrico

чт, 24 мар. 2022 г. в 18:34, Elaine Meng <meng at cgl.ucsf.edu>:
>
> Hello,
> I don't think it is possible to do this entire workflow in commands, because even if you do calculate some measure of similarity (like RMSD) with a command, there would still need to be detection of the highest similarity model and then retaining only that model.  I.e. it would require some kind of programming like python (beyond my skill set) if you are not going to look at these results interactively, e.g. look at the Log to figure interactively which one was the most similar.
>
> I will try to address command possibilities for measuring similarity, however.
>
> If the docking results are just lots of a different positions of the same molecule (although different from the crystal-structure ligand) then you could figure out specifically which atoms in the docked molecule should be paired with which atoms in the crystal-structure ligand, and then specify them *in order*  (so that correct pairing occurs) in "rmsd" commands to measure the RMSD.
> <https://rbvi.ucsf.edu/chimerax/docs/user/commands/rmsd.html>
>
> If the docking results are a bunch of different molecules, then the above will not be possible.  Instead you could compute centroid and axes for the crystal ligand and for each docked ligand (see command "define") and then measure distances between their centroids ("distance") and angles ("angle").  Or maybe you could even calculate planes and measure the angle between the planes of the two ligands, if that seems useful.  I imagine something like: calculate crystal-ligand centroid and axis once, then for each docked ligand, calculate its centroid and axis, do the measurements comparing to crystal ligand centroid and axis, delete docked-ligand centroid and axis, calculate centroid and axis for the next docked ligand, etc.  This would require you to experiment for a while entering the commands by hand to figure out exactly what you want to do and what measurements can really capture the type of similarity you have in mind.
>
> <https://rbvi.ucsf.edu/chimerax/docs/user/commands/define.html>
> <https://rbvi.ucsf.edu/chimerax/docs/user/commands/distance.html>
> <https://rbvi.ucsf.edu/chimerax/docs/user/commands/angle.html>
>
> I hope this helps,
> Elaine
> -----
> Elaine C. Meng, Ph.D.
> UCSF Chimera(X) team
> Department of Pharmaceutical Chemistry
> University of California, San Francisco
>
> > On Mar 24, 2022, at 8:04 AM, Enrico Martinez via ChimeraX-users <chimerax-users at cgl.ucsf.edu> wrote:
> >
> > Dear ChimeraX users!
> > First, I very much appreciate the support of the Chimera team related
> > to my questions!
> >
> > I have another question regarding analysis of the protein ligand
> > interactions in the docking poses saved in multi-model format. I need
> > to find a possibility (e.g. via some script that could be executed in
> > the chimeraX) to compare each docking pose with the X-ray structure
> > (loaded as the separate model) in order to select automatically the
> > model (=
> > docking solution) which may fit better to the X-ray structure. The
> > problem is that the ligand used in the docking is not the same as in
> > the X-ray structure (although they are quite similar in the functional
> > groups).
> >
> > Assuming that the both pdbs ((docking poses, and X-ray structure))
> > have been superimposed (based on the protein atoms) how could I
> > automatically switch to the model (in the docking ensemble) that has
> > the closest position of the ligand as in the X-ray structure?
> > Here is the example of the workflow:
> >
> > open docking_ensemble.pdb
> > open xray.pdb
> > # fit X-ray structure to the first docking solution based on the protein atoms
> > match #2 to #1.1
> > # compare each docking pose (ligand with X-ray and determine which
> > model has the most similarity to it
> > # determine the model #1.X which would have closest similarity to the
> > X-ray structure
> > # close all the models, with the exemption of the model #1.X
> > close ~ #1.X
> >
> > I would be grateful for any suggestions for possible algorithms!
> > With kind regards,
> > Enrico
>