The Molecular Graphics Laboratory Forum

Hi all,

I'm currently using Vina to and comparing the docking results from antagonists and agonists of a receptor. The docking of the agonists works just fine, but the antagonists are not docked in useful poses. This is weird, since the original receptor structure has been crystallized with an antagonist bound. My suspision is that Vina disfavours larger molecules, antagonists for this receptor tend to be larger and have more flexible bonds as I can say from the analysis of the docks of my database.

Is there a way to tweak the scoring function to enable e.g. soft repulsion? Or scale the vdw radii? Maybe just increasing the exhaustiveness could do the trick (larger comformational space to sample?)?.

I'll appreciate any suggestions .

Many thanks,
Michael

Not by the end users at present. When/if Vina is released as open-source, you should be able to change the weights yourself. However, frankly, I don't think this is the right thing to do in your case. Correct binding modes simply aren't guaranteed for every receptor-ligand pair by molecular docking. Ad hoc changes of the weights are unlikely to have beneficial effects in general.

_________________
Please do not email or PM me with questions. Vina.scripps.edu/questions.html explains where to find help.

You can try --exhaustiveness=100 to increase the exhaustiveness, but in my experience (and according to the paper) this doesn't help much.

Hi,

I have done some studies with autodock-vina and other molecular docking approaches. One result of these studies was, that the training of the scoring function is very important for a successfull screen. This means, when you are able to test/validate a molecular docking approach by comparison with experimental data, you can really benefit, when you optimize your scoring function for the particular target. E.g. weight the electrostatics a bit more, because these terms might be not equally important for all proteins.

Therefore, I strongly wanna suggest such a functionallity for autodock and/or autodock vina. I think the community as well as particular studies can really benefit from that! It should not be to much effort, right? To pronounce it a bit more, it should just be possible to change the weights of the scoring function. The particular tests and optimizations can be "outsourced". Whether the optimization is then done in terms of enrichment of (known) active compounds or by RMSD to known structures, could than be decided from case to case.

Thanks
Sören

I opened a new thread about that issue: viewtopic.php?f=12&t=1912&p=4998#p4983