Merge pull request #2752 from RosettaCommons/jadolfbr/ab_design_optimize_interface
MonteCarlo Interface
Main
===
This PR adds a new derived MonteCarlo class called `MonteCarloInterface`, located in `simple_moves`. It uses the `InterfaceAnalyzerMover` to score the interface and optimize the **dG** of the interface specified. **dG** and **total_energy** CAN be weighted, with default of dG at 1.0. This can be integrated into protocols at will. Due to the need of a specific interface, it is implemented as protocol-specific instead of within `MonteCarlo` itself.
The relevant option that can be accessed by a protocol to appropriately construct the `MonteCarloInterface` is `-mc_optimize_dG`. Options to change weights are given by `-mc_interface_weight` and `-mc_total_weight`
Etc.
===
- This has been integrated into the `RosettaAntibodyDesign` protocol (enabled using `-mc_optimize_dG`) and shows significant improvement in interface quality in resulting models. The models should still be pruned/filtered by total energy, only considering the top 10 or 25 percent of decoys.
- `RosettaAntibodyDesign` now minimizes using the neighbor list. I still don't know why this is not the default across Rosetta. Not using it is a pathology.
- Small bug fix in RAbD to only add atom-pair constraints by default if using docking. Otherwise, they should be turned off.
- Change the way cluster cutoffs work in RaBD to now be gene and cluster specific.
Unit Tests
=======
Add decently comprehensive tests for MonteCarlo(!!!) And MonteCarloInterface