[Chimera-users] Chimera/Modeller to add missing residues

Thu Nov 16 11:01:01 PST 2017

Hi Daipayan,
The results are directly from MODELLER, so although I do not know the detailed reasons, it produces small shifts in coordinates and chooses to write the atoms in that order.  The shifts appear to be rigid-body movements of the structure as a whole without changing the actual structure (internal coordinates) of the parts that are not being modeled.  In other words, if you start with chain A residues 1-10 and make models of chain A residues 1-13 with residues 1-9 kept the same and residue 10 allowed to move, even though the output coordinates of 1-9 are slightly different numbers, you can match them exactly onto the coordinates of 1-9 in the original structure.

If you want more detailed reasons you could try asking on the Modeller forum:
<https://salilab.org/modeller/discussion_forum.html>

As mentioned in my earlier message, the Chimera-Modeller interface only allows modeling 1 chain at a time, so the shifts are completely meaningless as they do not affect the (internal) coordinates of the structure.  You might be concerned, however, if you are putting several different modeled chains back together to make a multichain structure.  In this case, the shifts are very small, so you could choose to ignore them.  Again, it really depends what you are going to do with the resulting final model.  If you were really concerned about the small shifts, to get rid of them completely, you could match the model back onto the corresponding atoms of the original structure before saving the coordinates.

I tried making 5 models of 1YPH chain A adding the missing C-terminal residues.  When I save the models to PDB and then look at the coordinates with a text-editor, they all have similar small shifts of the N-terminal residues as you reported.  However, they are  barely perceptible when you view the structure. 

Here the original chain A is purple, the five models tan, without any matching.  You can barely see any difference except in the modeled C-terminal part on the right.  RMSDs of all 60 atoms of residues 1-9 of my five test models compared to the original were in the range 0.044-0.068.  (measured with commands like: rmsd #1:1-9.a #0.1:1-9   …in my case the original is #1 and the five models are #0.1-5)

Then I matched the models (#0.1-5)  onto the original (#1), commands:
match #1:1-9.a #0.5:1-9
match #1:1-9.a #0.4:1-9
match #1:1-9.a #0.3:1-9
match #1:1-9.a #0.2:1-9
match #1:1-9.a #0.1:1-9
… then all the RMSDs were zero (0.001 but that’s within coordinate rounding error).

I hope this helps,
Elaine
----------
Elaine C. Meng, Ph.D. 
UCSF Chimera(X) team
Department of Pharmaceutical Chemistry
University of California, San Francisco

> On Nov 15, 2017, at 7:27 PM, Daipayan Sarkar <sdaipayan at gmail.com> wrote:
> 
> Hi Elaine, 
> 
> Thanks for your detailed instructions. My apologies for the confusion but your guess about the different models is correct. After following your instructions, I compared the first residue of the original PDB file with the first residue of 'model 2 (residues missing in chain A)' and following is what I got. Note, the x, y, z coordinates of the atoms have changed and the sequence of atoms has changed. I am not sure why this happens. Can you help me understand? Note, the missing residues where added in model 2 but I am not showing here as there is nothing to compare against.
> 
> The coordinates for chain A of the first residue in original PDB file:
> ATOM      1  N   CYS A   1      32.449  -7.381  50.051  1.00 18.85           N
> ATOM      2  CA  CYS A   1      31.974  -7.012  48.720  1.00 17.01           C
> ATOM      3  C   CYS A   1      33.101  -6.917  47.728  1.00 15.67           C
> ATOM      4  O   CYS A   1      34.254  -6.671  48.116  1.00 16.39           O
> ATOM      5  CB  CYS A   1      31.253  -5.658  48.764  1.00 17.77           C
> ATOM      6  SG  CYS A   1      32.167  -4.263  49.459  1.00 19.16           S
> 
> 
> The coordinates for chain A of the first residue in PDB file after adding missing residues using Chimera-Modeller:
> ATOM      1  N   CYS     1      32.426  -7.418  50.092  1.00  0.00           N
> ATOM      2  CA  CYS     1      31.954  -7.042  48.762  1.00  0.00           C
> ATOM      3  CB  CYS     1      31.233  -5.687  48.812  1.00  0.00           C
> ATOM      4  SG  CYS     1      32.146  -4.297  49.517  1.00  0.00           S
> ATOM      5  C   CYS     1      33.082  -6.942  47.773  1.00  0.00           C
> ATOM      6  O   CYS     1      34.235  -6.698  48.164  1.00  0.00           O
> 
> 
> Regards,
> Daipayan
> 
> 
> On Tue, Nov 14, 2017 at 12:00 PM, Elaine Meng <meng at cgl.ucsf.edu> wrote:
> Forgot to finish this one thought, sorry...
> 
> > On Nov 14, 2017, at 9:47 AM, Elaine Meng <meng at cgl.ucsf.edu> wrote:
> >
> > An important point is that the Chimera-Modeller interface will only model one chain at a time.  It does not care that other chains are there, so any added residues/loops might end up clashing with the other chains.  In this case it didn’t seem
> 
> … that bad since the added residues were on the outside of the complex.  However, Find Clashes/Contacts (in menu under Tools… Structure Analysis) identified a few clashes between those chains and the other chains in the merged model in my test.  I only made one Modeller model of each chain.  One reason you might want to try making more than one model of each chain is that they may have different numbers of clashes with the other chains in the structure.  However, whether you care about this depends on what you are going to do with this overall structure. If it’s just for a picture, it’s not important, or if you were going to run minimization or dynamics, the clashes would easily resolve themselves in this case because they involve peptide termini on the outside of the complex.
> 
> Finally, if you want to use Modeller to model the multi-chain complex with awareness of the other chains, you would have to use it separately outside of Chimera.
> 
> Elaine

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://plato.cgl.ucsf.edu/pipermail/chimera-users/attachments/20171116/6d6b3e88/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 5models.png
Type: image/png
Size: 31122 bytes
Desc: not available
URL: <http://plato.cgl.ucsf.edu/pipermail/chimera-users/attachments/20171116/6d6b3e88/attachment.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 5modelsmatched.png
Type: image/png
Size: 30734 bytes
Desc: not available
URL: <http://plato.cgl.ucsf.edu/pipermail/chimera-users/attachments/20171116/6d6b3e88/attachment-0001.png>