You are here

Noob question about optimizing a pose

4 posts / 0 new
Last post
Noob question about optimizing a pose
#1

Hello, I am very new to PyRosetta, so this might not make a lot of sense. What I'm trying to do is minimise the energy of a given pose. I don't know much about this, so any tips about how to approach it and which instruments to use would be appreciated. Thank you.

Category: 
Post Situation: 
Fri, 2019-04-26 10:49
AleksanderK

Your question makes sense, and there's no shame in asking a newbie question!  First, there are two types of minimization.  You can find the nearest local minimum (which is probably pretty close to your starting structure), or you can search for the global minimum (which might be a completely different structure than your starting structure).

To find the nearest local minimum, the Rosetta minimizer is used.  You can invoke this through RosettaScripts or PyRosetta with the MinMover (https://www.rosettacommons.org/docs/latest/scripting_documentation/RosettaScripts/Movers/movers_pages/MinMover), or from the command-line with the minimize application (https://www.rosettacommons.org/demos/latest/tutorials/minimization/minimization).  A variation is to consider alternative packings of side-chains before minimizing, for which you can use the FastRelax mover in RosettaScripts or PyRosetta (https://www.rosettacommons.org/docs/latest/scripting_documentation/RosettaScripts/Movers/movers_pages/FastRelaxMover), or the relax application (https://www.rosettacommons.org/docs/latest/application_documentation/structure_prediction/relax).

Searching for the global energy minimum -- a.k.a. de novo structure prediction -- is of course a harder problem.  Rosetta's primary tool for that is the AbRelax application (https://www.rosettacommons.org/demos/latest/tutorials/denovo_structure_prediction/Denovo_structure_prediction).

As a final note, minimization is deterministic (you should get the same answer no matter how many times you run it, if the inputs are the same), so you need only run it once.  FastRelax uses a stochastic side-chain optimization algorithm and deterministic minimization, so running it a few times could give you slightly different results -- so it's probably worth doing 10 independent runs and picking the lowest-energy result.  Rosetta AbRelax is stochastic, and is searching a vast, vast space, so you probably want to run it thousands of times and pick the lowest-energy result.

Fri, 2019-04-26 11:40
vmulligan
Fri, 2019-04-26 16:59
ajasja

Thank you, that was very helpful

Fri, 2019-05-03 10:56
AleksanderK