As stated in AutoDock Vina manual, you can set sidechains on the grid to be flexible (http://vina.scripps.edu/manual.html). Sidechains of the grid will be maintained rigid, what should not do any harm to your docking assay. If you intend (and you should!) to run molecular dynamics, its first minimization protocol will take care of any spurious bonded error.

The ligand will change its conformation during the docking. If you can guarantee that bonds, angles and dihedrals are set correctly in your molecule, I don't see any use for minimization. Still, you probably want to check whether its conformation is acceptable, since many small molecules topologies are not well described in Amber force field (force field used by AutoDock Vina, if I am not mistaken).

Hope it helps!

Marcelo

Dear Sandeep,

I think minimization of the protein has not to be a compulsory method to apply before docking because the original cavity and protein conformation (e.g. active or inactive conformation, soluble…) or in may suffer major changes and almost disappear. Doing a minimization you are assuming that a theoretical model is more valid than the experimental one (X-Ray, NMR…). In this sense, if you want to explore some mobility aspects, you can define flexible residues of the protein and perform a flexible docking.

On the other hand, a minimization of the ligand makes more sense to optimize bond distances and angles mainly in a quantum chemistry level calculation.

All the best,

Miguel

Thank you Polêto, Bera and Romero-Cuevas. So I conclude that its better go with energy minimization of ligand and in case of protein as suggested, its better to proceed with the minimization of theoretical models than experimentally resolved structures. My query what kind of minimization algorithm is to be used for ligand EM. I have ChemBioOffice, which has MM2 energy minimization algorithm. Can I use it or should  or is there anything else to suggest.

Hi Telkar,

For what you want (i.e. suitable bonds and angles), any EM algorithm will do the trick. Many studies now has shown that small ligand bound energy is not at either local or global minimum. Remember that the ligand will 'accomodate' into your pocket, so its geometry will change, but not those parameters. With that in mind, you don't need a perfect conformation. You just need a good conformation to start. The rest will be taken care of in flexible docking engine.

All the best,

Marcelo

Sandeep Telkar

If you minimize a protein first then you are making an assumption that the force field which you have applied, describes the geometry of a protein better than the crystal structure does. The minimization may end up with bad geometry, especially when the forcefield employed are different (for PDB and X-ray diffraction). Moreover, minimization may also lead to the elimination of the crystal-packing artifacts.