[Chimera-users] Coloring ensemble atoms by distance from mean structure

[Elaine Meng]

Hi David,
If I understand correctly, Tom’s script calculates an average and then allows coloring ALL of the ensemble structures (all their atoms) by their respective distances from the average.

(A) What I’ve done in the  past is to take a single structure (which could be the average or any single member of the ensemble) and color it by the per-residue RMSD across the ensemble.  So the differences would be: coloring a single structure, and by per-residue values only, rather than by per-atom.  It can also work for superimposed structures that are not strictly identical (not a true ensemble) as long as they are similar enough to be sequence-aligned.  Although this process doesn’t require a python script, it is a bit more involved since there are several steps, as outlined in this old post:
<http://plato.cgl.ucsf.edu/pipermail/chimera-users/2014-March/009710.html>

(note the process for associating an entire ensemble has been improved since then, and backbone and all-atom RMSDs per residue are available in addition to the CA RMSD mentioned in that post; however caveat emptor on the all-atom RMSDs as there are no corrections for symmetrical sidechains Phe, Tyr, … only atom names are used)

(B) Chimera does have an “average structure” calculation, although it may be simpler than what you want.  It can work on multiple PDBs or a multi-model single PDB as long as you open the ensemble as a trajectory (start Tools… MD/Ensemble Analysis… MD Movie, then in that dialog choose the PDB file input type).  Then in the MD Movie dialog’s menu, choose “Analysis… Average Structure”.  Limitation is that is can’t be used on a heterogeneous set of models, only a true ensemble where each member has exactly the same atoms with the same names, residue numbering, etc.
<http://www.rbvi.ucsf.edu/chimera/docs/ContributedSoftware/movie/framemovie.html>

Just mentioning this stuff because it seems related, although Tom G may have already addressed your needs more precisely.  :-)
Best,
Elaine
-----
Elaine C. Meng, Ph.D.                       
UCSF Chimera(X) team
Department of Pharmaceutical Chemistry
University of California, San Francisco


> On May 3, 2018, at 12:02 PM, Tom Goddard <goddard at sonic.net> wrote:
> 
> Hi David,
> 
>  I’ve attached a Python script mean_distance.py that can compute the distance of each atom in an ensemble from the mean structure.  Then you can color using that distance.  Here’s an example set of Chimera commands.
> 
> open 1mtx
> show
> open mean_distance.py
> rangecolor distance_from_mean 0 blue 3 white 6 red
> 
> The script uses all open molecules as the ensemble.  The coloring can also be done with menu Tools / Depiction / Render by Attribute.
> 
> 	Tom
> 
> <mean_distance.py>
> 
> <1mtx_mean.png>
> 
>> On May 3, 2018, at 10:10 AM, Morgan, David Gene wrote:
>> 
>> Tom,
>> 
>>   Hope you are doing well.  Jim Powers and I here at IU keep thinking we need to invite you and/or the team back for another chimera workshop, but we haven't done the ground work to see how possible that would really be.  But I keep hoping!
>> 
>>   In the meantime, I have a question about whether chimera can do something, and if not, I have a feature request...  I was watching a grad student give a talk showing lots of superposed atomic structures.  The structures included some molecular dynamics simulations and some crystal structures of the same protein in subtly different conformations (packing into a viral capsid, where the different subunits in the asymmetric unit have slightly different conformations).  It occurred to me that _after_ showing the mess of spaghetti he had, the points he needed to make would be better served with a different visual:  If he were to average the ensemble of MD atomic positions to a single chain (and color-code by the ensemble RMS difference from the average) and do the same for all his ensemble of X-ray structures, it would be much easier to talk about and show the consistencies and inconsistencies in the atomic co-ordinates from all the data.  This same sort of averaging an ensemble of atomic co-ordinates is sometimes done with NMR structure ensembles (but I don't know about the color coding with the RMS).
>> 
>>   I have looked at lots of packages that manipulate atomic co-ordinates, but haven't found anything that seems to actually do what I want to do.  For example, gromacs is supposed to do the averaging but it really wants to work with MD data and at a minimum, I'm not smart enough to make it work with a series of standard PDB files.  My NMR colleague suggested molmol for the averaging, and I need to explore that more.  But there doesn't seem to be a recent linux version nor does it seem to be "scriptable" (i.e., if I wanted to do this for a dozen different ensembles, I would have to painfully mouse click my way through the process).  On top of all that, I ran into numerous on-line discussions about how averaging was "the wrong thing to do."  All those people had valid reasons for saying so, but right now, I'm really talking about a more simple visual that conveys some of the information contained in any sort of ensemble of atomic co-ordinates.
>> 
>>   So my first question is:  does chimera have a utility that would do this sort of thing?  I looked at various commands but did not see anything that looked like it would do this.  If not, can I put a feature request into the queue?
>> 
>>   Looking forward to hearing from you and chatting some more.
>> 
>> --
>>   politics is more difficult than physics.
>>                                            A. Einstein
>> 
>>           David Gene Morgan
>>       Electron Microscopy Center
>>            047D Simon Hall
>>            IU Bloomington
>>         812 856 1457 (office)
>>         812 856 3221 (3200)
>>     http://iubemcenter.indiana.edu
>> 
> 
> _______________________________________________
> Chimera-users mailing list: Chimera-users at cgl.ucsf.edu
> Manage subscription: http://plato.cgl.ucsf.edu/mailman/listinfo/chimera-users