An overview of this energy function optimization effort is available here.

For information on the previous set of updates to this score function, see the beta nov16 score function page here.

For most protocols (those that use getScoreFunction to set the protocol score function), the flag -gen_potential or simply -beta (which will always load the latest beta energy function) will load this version of the beta energy function.

For RosettaScripts protocols, the flag -gen_potential or -beta must be provided, and the following scorefunction declaration must be made:

<ScoreFunction name="beta" weights="beta_genpot"/> or <ScoreFunction name="beta" weights="beta"/>

New atom typing

An alternate atom typing scheme has been introduced. An alternate param file generation app, scripts/python/public/generic_potential/ has been added to handle parameter generation with the new atom typing. NOTE: this new typing is completely separate from protein atom typing. Fore more information how thse atom types are defined and how this works, see here.

  • A generic torsional potential, gen_bonded has been added using these new types. This potential is undefined (returning 0) for non-ligand residues.
  • LK and LJ parameters have been fit for these new atom types using a combination of small molecule crystal data and ligand-bound protein structures.

Generalized torsion term

A generalized torsion term called "gen_bonded", a component of the generalized energy function, is defined according to the atom types and bond types at the torsion axis. It uses the common functional form used in regular force field, representing torsional preference as a series of cosine functions up to 4-th order. The parameters associated with the term can be found from the file located at the Rosetta database: database/scoring/score_functions/generic_potential/generic_bonded.round6p.txt

Ligand docking

A new ligand docking protocol has been added, GALigandDock, that takes advantage of this new generic potential. It combines fast scoring on a precomputed grid with a genetic algorithm to allow very accurate ligand docking in 3-10 CPU minutes total per target (fixed sidechain) or 10-30 CPU minutes total (flexible sidechain).

Ligand preparation

Whole process preparing for GALigandDock runs starting from SMILES string or a mol2 file and assigning proper Generalized atom types through can be found here.