I ran a MD using Gromacs but this is my first time doing ensemble docking. How can one extract representatve structures from the trajectory?
I followed a tutorial that used gmx cluster to the centroid conformation from each calculated clusters, but are these ideal structures to work with?
yes it can be done by gmx cluster. you may use gromos clustering, use of reference structure is also optional. you may consider average rmsd of the trajectory as cutoff and take the middle most structure of the largest cluster as representative structure.
Maybe the simplest way is to same structures with a certain interval,
which will allow unbiased sampling of free energy surface, at least in theoretically.
Clustering based on rmsd seems to be good in the aspect of sampling diverse structures.
If you are interested in docking small ligands, I think that limiting rmsd calculation at the binding pocket might be more helpful as the local pocket structure have larger impact on the docking simulation.
I'd suggest that after selecting your cluster centers (which is a reasonable thing to do) you energy-minimize them and then check if some of the structures did not converge to same local minima.
You can also calculate the interaction patterns (vdW, HB, etc) of each ligand structure with the protein and use these patterns for clustering and subsequent analysis.
Thank you everybody. Your answers were really helpful. I think i'm on the right way now!
During surveying recent researched, I found an article that can help you.
Offutt,T.L. et al. (2016) Enhancing Virtual Screening Performance of Protein Kinases with Molecular Dynamics Simulations. J. Chem. Inf. Model., acs.jcim.6b00261.
This article suggest clustering with CA atoms around the ligand binding sites.
- Badges
- Science topic