ADL: Which conformation is the "true conformation"?

Mark Swingle mswingle at jaguar1.usouthal.edu
Thu Apr 23 16:28:28 PDT 2009


On Thursday 23 April 2009 10:30:55 am mcmartin at student.matnat.uio.no wrote:
> Hi,
>
> I'm fully aware of the limitations of docking software and that it should
> only be used as some kind of guidance prior to in vitro experiments.
> Having said that, I'm trying to understand the results for several
> compounds in which each docking gave me an .dlg output file with 200
> different conformations ranged by binding energy values and dissociation
> constants. So my question is, do you guys just pick the conformation of
> lowest energy and work further upon that? Or do you look upon the
> clusterings and see the number of conformations per energy value? Or the
> RMSD values?
>
I was just looking at a recent paper that might address your concerns:

http://www.ncbi.nlm.nih.gov/pubmed/18351616

http://www3.interscience.wiley.com/cgi-bin/fulltext/117943606/HTMLSTART

One interesting observation from their test cases was that random selection of 
a conformation (from the set of LGA solutions produced by autodock) was, on 
average, more successful (in terms of RMSD relative to the experimental 
conformation) than picking the lowest energy conformation. Lowest-energy was 
correct for 3/22 complexes and random had an expectation of 12.35/22 cases. 
Now, they were looking at weakly interacting complexes so the really poor 
performance of the scoring function here is probably not generally 
applicable. 

OTOH, using the scoring function but applying a correction factor based on the 
cluster populations was successful in 16/22 cases. This is probably your best 
bet if you don't want to do consensus scoring.

Regards,

Mark


More information about the autodock mailing list