[Chimera-users] different docking scores with SDFs and SMILES

[Elaine Meng]

Hi Kelly,
Thanks for the kind words about Chimera.

Personally I would not pay attention to the energy values, as long as the structures appear to be reasonable, not warped or strained.  Even if a molecule doesn’t have any rotatable bonds it is completely expected for the same compound built with different methods (or even just rotated slightly in 3D so that they look exactly the same when superimposed) to  have different molecular mechanics energies both before and after minimization, and different docking results. There are tiny roundings in coordinates, etc.  It may not even be necessary to minimize the structures, since some of the structure-generation processes already have minimization, and the minimization is all gas phase and not in the context of the binding site anyway.  If you are only doing rigid docking, the results will always be quite limited amongst the universe of reasonable possibilities.  If you are using the Autodock Vina interface in Chimera, you will be further restricted by the small amount of sampling available via that web service.

Nevertheless if you are concerned, look up the details of how the 3D compounds are built to try to decide which approach is better. 

If you enter a PubChem CID in Chimera it will fetch the SDF from PubChem.  How the compounds in “PubChem 3D” are  calculated is described in this page (you have to scroll down past the table of contents):
<https://pubchem.ncbi.nlm.nih.gov/release3d.html>

If you enter a SMILES in Chimera it will use a SMILES to 3D translator, either smi23d or the one provided by the National Cancer Institute.  Use “help open” command in Chimera to see more details on which is used and links to follow to get further information (in the “open” help page, go to the “files from the Web” section and look for SMILES).

I have no idea about the effects of using alternative SMILES strings for a compound, sorry. 

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

> On Jun 22, 2018, at 9:49 AM, Kerr, Kelly E (HSC) <Kelly-Kerr at ouhsc.edu> wrote:
> 
> Hello,
> 
> I have really enjoyed using your incredibly comprehensive and user-friendly program. I cannot overstate how much I appreciate it. I'm encountering a discrepancy though that I can't explain. It probably shows a lack of basic understanding on my part, but I would be grateful if you could help me figure this out.
> 
> I have not been able to find 3D structure files for all of the ligands I'd like to run, so I have been using Chimera to build structures from the canonical SMILES presented on PubChem. I have also been using Chimera to minimize these structures. In order to be sure that the structures are completely minimized I've used the following...
> Steepest descent steps	1000
> Steepest descent  step size	0.02
> Conjugate gradient steps	10000
> Conjugate  gradient step size	0.002
> 
> I ran this minimization and then AutoDock Vina for both the SDF and SMILES-built versions of several ligands. Both versions came from the same page (I was using PubChem) and both versions seemed to get minimized completely (the minimization outputs ended with many rows of RMSDs of zero).  I'm confused that I'm getting different energy values (both initial and after minimization) for the two versions and I'm getting slightly different docking results as well.
> 	• Is it expected that SMILES and SDF files of the same chemical should have different energy values even after minimization?
> 	• Would you recommend using Zinc (aromatic) SMILES to build structures instead of the canonical smiles on PubChem? OR would you recommend using PubChem CIDs instead of SMILES?
> Thank you so much in advance for your time and guidance.
> 
> Best regards, 
> Kelly Kerr