I am performing docking using autodock.
In the procedure, am giving 35 GA runs.
Normally we will select the first one from the 35 docked position which has the least binding energy.
In my case, First docked structure is binding at some other position in the grid other than required place.
But second, third and fourth are binding exactly where i want.
My question is, can I consider the other docked structures out of these 35 runs and their binding energy as a result of my docked structure?
Hi Sajin,
Yes you can consider other docked poses also. It is not a strict criteria that you have to select the least binding energy pose. You can try with other docking software like Dock6 and SwissDock to verify the binding pose that you are going to consider.
Yup, if you know which residues are important for binding then you can analyze the different poses which are showing interaction with the residues of interest .
Dear Sajin,
You can surly select the second, third and four conformation. You can also analyze the number of interacting residues, hydrogen bonding and binding energy (kcal/mol) and cluster RMS as well. Also, you can compare these features with your first structures. I think you need to set the grid box very carefully before the creation of the dpf file. Anyway your second, third and fourth conformation will give you some insights.
No one should provide "confirmation" based on any simulated docking result. Anything akin to confirmation comes from primary sources, most commonly x-ray crystallography. Yet I would suppose that a key factor of your question is found in: "But second, third and fourth are binding exactly where i want."; namely, "where [you] want". This forces the meta question: what do you want? what is the goal of your investigation? I'm sure it's not merely to have your initial hypothesis "confirmed". But perhaps there are other factors to "want" alongside the discovery of the levels of uncertainty about the system in reality? For the most part, I've arrived at the opinion that virtual docking is most powerful to eliminate what is notably unlikely.
Jeffrey Godden
Hi sir,
Thank you for your valuable answer.
I didn't mean that it is confirmation. Just an hypothesis like my test ligand may inhibit the natural substrate for a protein or enzyme.
Secondly, Where i want is nothing but the active site. Some of my test ligands are not docked into active site. Rather than binding to the active site it binds to place near the active site.( May be its because of grid parameters i selected for docking).
Regarding my work am trying to inhibit the fusion of viral proteins with the host cells.
with regards
Sajin
I don't see any problem selecting the second, third or fourth listed pose as Autodock's scoring function, as any other scoring function, is not perfect. In my opinion, it is also a matter of consistency, do you consistently observe the binding modes listed in second or third position? Have you thought about rescoring your list of poses with another scoring function (another one from Autodock, DrugScore, ChemScore, ITScore, MM/GBSA, etc.). Using multiple independent evaluation methods that consistently output similar results will make your docking experiment way more robust.
Dear Sajin,
You are facing a very usual situation of anyone performing docking, which is a modeling technique including a bench of approximations because meant for rapid description of ligand/protein binding. Scoring in particular are very simplified functions to roughly evaluate the goodness of fit (and not binding energy as sometimes found in the literature). Echoing Louis-Phillipe, a good way to gain confidence is to use a consensus approach. For instance by using multiple engine/program for the same problem. Or simpler, by restoring all solutions given by AutoDock (important is to use as orthogonal as possible techniques, many free-for-academics options are listed at www.click2drug.org). You may also want to estimate the binding free energy. For this latter, a broad range of molecular mechanics techniques do exist. Should you want to go simple first, the swissdock server (www.swissdock.ch) gives you very simple access to a docking methodology, that evaluates docking poses through an energy calculation using the entire CHARMM force field.
Many options, certainly problem-dependant choice...
Good luck!
Antoine.
Hello
you may want to look at the number of conformations in each cluster, which you can obtain from the dlg file. The cluster with the highest number of conformations can be considered as your best binding mode, since the program docked the ligand there most frequently. Choosing the cluster with the lowest binding energy seems to be rather questionable.
Fur further info, please refer to the following publication: http://pubs.acs.org/doi/abs/10.1021/bi300299z
best Luck!
Maen
- Badges
- Science topic
- Similar topics
- Mathematics