Hello,
I want to dock on major representative forms of an enzyme using "snapshots" from trajectory. I ran an 8ns MD simulation for a holo-enzyme in order to account for changes in the binding site but I dont know how to select those representative structures of the enzyme pocket.
The most intuitive option is to cluster them according to RMSD. But should i superimpose the main-chain atoms then calculate RMSD for the pocket ,N-terminal then RMSD ,or superimpose the pocket c-alpha atoms ??
Working on Discovery studio 2016
Hello,
I should superimpose the backbone of the enzyme and then, If a RMSD value is the same for two different structures (it should not happens), I calculate the RMSD of the side chains of the active site residues. Every structure group is defined by these two variables.
I hope it helps you
Hey there,
I don't think you should trivialize this. A better way would be to use all representations and expose the ligand to all possible shapes of the pocket on the protein. You should then minimize the ligand while it is inside the pocket (not more than 200-300 steps of a good minimizer like conjugate gradient) and follow that by a short simulation. Energy calculations over the short simulation should be taken. Once you have the energy value -- you should cluster those.
The reason why I say that this is better - is because when you cluster you take an average pocket to represent the entire cluster. This is not a good idea because you may miss the ideal pocket. Using all pockets and then clustering based on energy sounds to me as a better approach.
However if you are intent on clustering the pockets based on RMSD - since the structures are identical "YOU SHOULD" get the same clusters (not same RMSD but same clusters) irrespective of your choice of atoms to superimpose.
Hope this helps.
Best,
/A
Thanks a lot !!
Any suggestion for useful software to do cluster analysis ?
- Badges
- Science topic