This energy function optimization aims to reparameterize the Rosetta energy function subject to several different constraints:
These constraints are encoded as an "energy function goodness" target function and derivative free optimization of a relevant subset of energy function parameters is performed, maximizing this function. The relative weights of each of the criteria listed above are selected in an ad hoc manner; ideally, weights are chosen such that no individual test worsens following optimization.
Many rounds of optimization have been performed, optimizing different subsets of energy function parameters. Periodically, these optimized parameter sets are tested on independent data. Large improvements are added as a flag to Rosetta.
These flags include (more details of specific improvements can be seen by following the links below):
For brief explanation of new terms (e.g. rama_prepro and so on), follow this link
For most protocols (those that use getScoreFunction to set the protocol score function), the flag -beta will load the latest version of the beta energy function.
For RosettaScripts protocols, the flag -beta must be provided, and the following scorefunction declaration must be made:
<beta weights=beta/>
Additionally, variants of the scorefunction may be specified through RosettaScripts or through the flag -score::weights: