Energy Minimization Result Problem

Question

[nehabharty123]

Hello

I am new user of gromacs. I modeled a structure of protein with the help of Homology modeling and after that I validate the structure using Ramachandran plot with the help of procheck server. The residues present in most-favored region was 89%. To refine the model I tried to perform energy minimization and MD Using gromacs. But after energy minimization when I check my result again in procheck server the ramachandran plot become become worse and the residues present in most-favored region decreases to 87%. after MD also the result was same.

As I am new user of gromacs I don't know that is this a normal that after energy minimization result can be decreases??

Also can I use this energy minimize structure for docking?? or should I use the previous structure which is having 89% residues in most favored region.

The detail of energy minimization is as follows:

Force Field Used: CHARMM27 all-atom force field Water Model: TIP3P

I have also perform EM and MD with different force fields like AMBER, OPLS but the result was better in CHARMM27.

My em.mdp file is:

; VARIOUS PREPROCESSING OPTIONS = title = cpp = /lib/cpp include = define = -DFLEXIBLE

; RUN CONTROL PARAMETERS = integrator = steep ; start time and timestep in ps = tinit = 0 dt = 0.002 nsteps = 15000

; ENERGY MINIMIZATION OPTIONS = emtol = 10 emstep = 0.1 nstcgsteep = 1000

coulombtype =pme nstenergy =10

It would be great if any one will help me out with this.

[nehabharty123]

Hello Joao

Thanks for your sujestion. The % identity between my sequence and your template is 46%.

I am using that structure for virtual screening. I am doing virtual screening with Homology modeled structure and docking final docking with selected small molecules with refined structure.

Thanks

Answer 1

[João Rodrigues]

Energy minimization is anyway performed after modelling (usually). If you want to use your model for docking, then I wouldn't bother much with the optimization. If you can, and it depends on the docking goal, use multiple input models, so multiple snapshots of the model from the MD (don't bother with the ramachandran plot results so much) to have a good sampling of your side-chains.

The most important thing of your modelling is anyway the % identity between your sequence and your template. If this is below 30%, you are in a 'danger' zone and refinement will rarely help you. Above that threshold, refinement might bring you closer to the native state, but since you want to perform docking, your 'native' state is not so important as it would be the unbound (apo) conformation.