flexpep docking, select the decoy to do the further docking, questions

[Anonymous]

Dear all:
I performed the flexpep docking by first prepack the interface. Then I conduct the docking with command:
flexpep -database rosetta_database/ -s prepacked.pdb -native prepacked.pdb -out:file:silent decoys.silent -out:file:silent_struct_type binary
-rbMCM -torsionsMCM -ex1 -ex2aro -use_input_sc -nstruct 100 -lowres_preoptimize
after that, I sorted the score value and select the top 5 decoys. (Is this the common way to select the decoys?)
But How can I further dock the decoys to refine the conformations? ( Do we usually refine the initial docking conformations?)
Should I perform the docking without the -lowres_preoptimize parameter for thse top 5 decoys?
In the manual, this is not stated clearly.
Hoping someone can tell me the right step to do this.
Thanks !

Michael

[Anonymous]

Dear Michael,
The number of structures to create, lowres pre-optimization and, iterative docking (re-doing docking on the top x decoys) all depend on your system, and your estimate on how close your initial structure is to the native.
In our original benchmark, we do: 100 models with -lowres_preoptimize, and 100 w/o, and select the top 10 and that’s it – i.e. no further refinement (by the way, indeed sorting according to the score is the way to go) You do not need to dock these 10 again, however, we did show that re-docking these shows modest improvement. So basically it’s a question of how much time do you have.
Re. -lowres_preoptimize, this is but a step in the overall protocol followed by high resolution refinement in full-atom mode, so the models you have are still full-atom even if you used this flag. The main difference between running with and without this flag is the possible range of sampling, with the flag you are sometimes able to sample for divergent conformations.

Is it clearer now?
Cheers,
Nir.

[Anonymous]

Dear Nir:
Thank you so much! You comments are very helpful.

Best wishes,

Michael

[Author]

Posts