From owner-chemistry ":at:" ccl.net Thu Feb 8 14:26:00 2007 From: "Rene Thomsen rt**molegro.com" To: CCL Subject: CCL: Molegro Message-Id: <-33562-070208135609-14136-TB9V2vmzotPa2UZSNkMVPA|a|server.ccl.net> X-Original-From: "Rene Thomsen" Content-Disposition: inline Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=ISO-8859-1; format=flowed Date: Thu, 8 Feb 2007 18:54:43 +0100 MIME-Version: 1.0 Sent to CCL by: "Rene Thomsen" [rt!^!molegro.com] Hi all, For protein-ligand docking, we would consider a hexapeptide to be a big molecule (i.e. more than 15 flexible torsions). Since our docking procedure (which is build on genetic algorithms) is stochastic, different runs may find different energy minima. For simple problems (small ligands) a single run is usually enough. For more complex problems you should run several runs (20 or more), use the clustering options, and e.g return the five top-ranked poses. Ideally, the highest scoring pose represents the real binding mode of the molecule, but sometimes manual inspection can improve the pose ranking. A couple of suggestions: - Focus the search - make sure the search space you consider is not too large. - Consider constraining the search if you suspect a certain interaction to be present. - Check the preparation of the complex (protonation state, missing atoms, check for warning/errors in the docking wizard, ...). Regarding the cavities: our cavity detection algorithm is also stochastic, so cavity sizes may differ. However, if you get different cavities each time you run the cavity detection algorithm, there is probably no well-defined binding pocket - which will make it more difficult for the program to succeed. You are also welcome to mail us your structures, so that we can have a look at your data to see if anything seems wrong. Kind regards, Rene Thomsen --- Molegro Hoegh-Guldbergs Gade 10, Bldg. 1090 DK-8000 Aarhus C Denmark www.molegro.com On 2/7/07, Richard Leo Wood rwoodphd(a)msn.com wrote: > > Sent to CCL by: "Richard Leo Wood" [rwoodphd!A!msn.com] > Hi all, > > My ligands are hexapeptides, so they are not really proteins, but I suppose could be classified as "small molecules". They are known to inhibit the receptor I am trying to "dock" them to (actually I'm trying to estimate the binding energies to the receptor). > > I have a concern about the results that I am getting. They don't seem to be reproducible. That is, if I run the same docking calculation over and over again, I get different results. I wonder why that is. I've noticed that the cavity sizes (for the largest cavity) vary from run to run, as well. > > I would think that if I used the same settings each time, I should get the same binding affinities, but I'm not. > > TIA, > Richard