I did this recently for an analysis of the subunit variation in AcrB. I took the LSQMAN multi-RMSDs among the three chains after improved superimposition and edited them into the format that Rob Campbell's data2bfactor.py expected. And then I put them into the B-factor column of one subunit. Finally I invoked Rob Campbell's color_b.py in pymol to colour the molecular surface by them (Fig. 4 in http://www.pnas.org/content/106/17/7173/suppl/DCSupplemental). But I guess a backbone could be similarly.
There are probably other ways to do this but Rob Campbell's scripts are a boon and a blessing....
cheers,
Martyn
Martyn Symmons
MRC-MBU
----- Original Message ----
From: Gerard DVD Kleywegt <[log in to unmask]>
To: [log in to unmask]
Sent: Tuesday, 23 February, 2010 14:57:02
Subject: Re: [ccp4bb] Per-residue RMSD for multiple structures?
Thanks Stephen!
I was going to suggest that, but I was afraid of the self-appointed CCP4BB Gestapo that has been seen goose-stepping in this neighbourhood recently (Tassos recently accused me of becoming mellow and diplomatic in my dotage, so I hope I've set the record straight now). However, since this solution is neither CCP4 nor Phenix, we may get away with this heinous act of bulletin-board heresy... On the other hand, I've learned that it is often more expedient to beg for forgiveness than to ask for permission.
I would add that:
- I assume that the sequences and numbering are identical
- you should put the structures in one big PDB file and read it into LSQMAN
- since LSQMAN doesn't do true multiple-structure alignment, you could pre-align them, e.g. with SSM/PDBeFold
- if you didn't, you could indeed use the MCentral and MAlign commands to align them
- my favourite plot would be the "CD plot" (but then again, it would, wouldn't it?) - see for instance: http://xray.bmc.uu.se/cgi-bin/gerard/image_page.pl?image=usf/pics/cdplot_1ldn.gif - which is also produced with the MPlot command - http://xray.bmc.uu.se/usf/lsqman_man.html#S82 - a normal MPlot would look like this: http://xray.bmc.uu.se/cgi-bin/gerard/image_page.pl?image=usf/pics/mplot_1ldn.gif
- the output file of the normal MPlot command is in a form that can be quickly converted into an O datablock for those handy with an editor and familiar with O datablocks, and could then be used to ramp a model inside O
- you may also want to consider showing how the (main-chain or side-chain) torsion angles differ between the structures, e.g. by plotting the circular variance of phi and psi - see for instance http://xray.bmc.uu.se/cgi-bin/gerard/image_page.pl?image=usf/pics/vmain_1ldn.gif - as described here: http://xray.bmc.uu.se/usf/lsqman_man.html#S83 - or a multiple-model Ramachandran plot like this http://xray.bmc.uu.se/cgi-bin/gerard/image_page.pl?image=usf/pics/mrama_1ldn.gif (with the MRama command). The advantage is that no superposition is required at all and that any domain movements won't debeautify your results
--dvd
On Tue, 23 Feb 2010, Stephen Graham wrote:
> I am pretty sure you can do this using LSQMAN from Gerard (BluRay?) Kleywegt.
>
> The pertinent commands are MCENTRAL to determine the 'most
> representative structure' (i.e. the one to align upon and show in the
> figure), MALIGN to do the alignment and then MPLOT to calculate a
> 'multi-RMSD' for each residue (see manual for details - set the
> 'cut-off for printing' to 0 to get all values).
>
> Regards depiction, I think pymol can also represent structures as
> sausages based on their B values:
> cartoon putty
> show cartoon
>
> HTH,
>
> Stephen
>
> On 23 February 2010 01:31, Ethan Merritt <[log in to unmask]> wrote:
>> Hi all,
>>
>> I am comparing 4 very similar (<1.5A rmsd) large (750 residues) structures,
>> but struggling to find a way to generate a figure that conveys where they
>> are most alike and where they diverge.
>>
>> Simply drawing a superimposed set of backbone traces results in what looks
>> like colored spaghetti. I don't think that's going to work.
>>
>> So I had the idea of drawing a single backbone trace, or ribbon diagram,
>> and coloring by the RMSD of the four C-alphas at each residue position.
>> But I can't find a program that will output this as a table of numbers
>> I can use. All of the multiple structure superposition programs must
>> have this information internally. After all, that's what they are minimizing.
>> But do any of the programs provide an option to write it out?
>>
>> I can get pairwise per-residue deviations by doing SSM superposition in Coot,
>> but that doesn't get me to an RMSD for all four structures jointly.
>>
>> Ethan
>>
>>
>> --
>> Ethan A Merritt
>> Biomolecular Structure Center
>> University of Washington, Seattle 98195-7742
>>
>
>
>
> -- Dr Stephen Graham
> 1851 Research Fellow
> Cambridge Institute for Medical Research
> Wellcome Trust/MRC Building
> Addenbrooke's Hospital, Hills Road
> Cambridge, CB2 0XY, UK
> Phone: +44 1223 762 638
>
Best wishes,
--Gerard
******************************************************************
Gerard J. Kleywegt
Dept. of Cell & Molecular Biology University of Uppsala
Biomedical Centre Box 596
SE-751 24 Uppsala SWEDEN
http://xray.bmc.uu.se/gerard/ mailto:[log in to unmask]
******************************************************************
The opinions in this message are fictional. Any similarity
to actual opinions, living or dead, is purely coincidental.
******************************************************************
|