Return to RosettaScripts

Overview

GeneralizedKIC selectors (class protocols::GeneralizedKIC::GeneralizedKICselector) choose a single solution, based on some criterion, from the many solutions produced by GeneralizedKIC. A single attempt at closure with GeneralizedKIC can produce 0 to 16 solutions, and one typically attempts many closures, so it is necessary to specify a method of choosing a single solution.

Use within RosettaScripts

Each GeneralizedKIC mover has one and only one GeneralizedKIC selector assigned to it. In RosettaScripts, this must be specified in the <GeneralizedKIC> block as follows:

<MOVERS>
...
     <GeneralizedKIC ... selector="&string" selector_scorefunction="&string" selector_kbt="(1.0 &real)" pre_selection_mover="&string">
          ...
     </GeneralizedKIC>
...
</MOVERS>

The selector_scorefunction and selector_kbt tags are optional, and are only used by certain selectors.

Types of selectors

  1. Random (selector="random_selector")
    As the name implies, this selector randomly chooses a solution from those found. No options are considered.

  2. Lowest energy (selector="lowest_energy_selector")
    This selector chooses the lowest-energy solution found. It should be provided with a scorefunction (selector_scorefunction tag). Since no repack is performed before scoring, it is recommended to use a scorefunction that only considers backbone geometry (see below). An example would be a scorefunction consisting of some combination of the hbond_lr_bb, hbond_sr_bb, rama, omega, and p_aa_pp terms. The selector_scorefunction tag is the only option considered by this selector.

  3. Boltzmann-weighted random (selector="boltzmann_energy_selector")
    This selector chooses a solution randomly, but weighted by a factor of exp(-E/kbt). This means that it must score the energy of each solution, so as with the lowest energy selector, this selector must be provided with a (preferably backbone-only) scorefunction with the selector_scorefunction tag. It must also be provided with a Boltzmann temperature, in Rosetta energy units, with the selector_kbt tag. The selector_scorefunction and selector_kbt tags are the only options considered by this selector.

  4. Lowest RMSD (selector="lowest_rmsd_selector")
    This selector chooses the solution with the lowest Cartesian-space RMSD to the input loop conformation. It is therefore best used on structures that already have a closed loop conformation that you would like to perturb subtly (e.g. with the perturb_dihedral GeneralizedKIC perturber).

  5. Lowest delta torsion (selector="lowest_delta_torsion_selector")
    This selector chooses the solution with the lowest torsion-space RMSD to the input loop conformation. Like the lowest_rmsd_selector, it is best used on structures that already have a closed loop conformation that you would like to perturb subtly.

Recommended scorefunctions for energy-based selectors

Energy-based selectors, such as the lowest energy and Boltzmann-weighted random selectors, must score candidate solutions with some scorefunction. The full-atom talaris2013 scorefunction would generally be a poor choice for this, because no repacking or energy-minimization is carried out before scoring to eliminate side-chain clashes. For this reason, a scorefunction that either considers only backbone geometry, or one that considers backbone and side-chain geometry but not long-range side-chain interactions, is recommended. Two examples are given below.

<bb_hbond_tors weights="empty.wts" symmetric="0">
#A backbone-only scorefunction that only includes the backbone hydrogen-bonding, omega, rama,
#and p_aa_pp terms from the talaris2013 scorefunction.
	<Reweight scoretype="hbond_sr_bb" weight="1.17" />
	<Reweight scoretype="hbond_lr_bb" weight="1.17" />
	<Reweight scoretype="omega" weight="0.5" />
	<Reweight scoretype="rama" weight="0.2" />
	<Reweight scoretype="p_aa_pp" weight="0.32" />
</bb_hbond_tors>
<bb_hbond_tors_fadun_cst weights="empty.wts" symmetric="0">
#A scorefunction that only includes the backbone hydrogen-bonding, omega, rama,
#and p_aa_pp terms from the talaris2013 scorefunction, as well as the side-chain
#fa_dun term and constraint weights.  Note that side-chains are not repacked, so
#the fa_dun term will be identical for all side-chains EXCEPT those that are part
#of the loop to be closed (e.g. the side-chains of the cysteines of a disulfide).
	<Reweight scoretype="hbond_sr_bb" weight="1.17" />
	<Reweight scoretype="hbond_lr_bb" weight="1.17" />
	<Reweight scoretype="omega" weight="0.5" />
	<Reweight scoretype="rama" weight="0.2" />
	<Reweight scoretype="p_aa_pp" weight="0.32" />
	<Reweight scoretype="fa_dun" weight="0.56" />
	<Reweight scoretype="atom_pair_constraint" weight="1.0" />
	<Reweight scoretype="angle_constraint" weight="1.0" />
	<Reweight scoretype="dihedral_constraint" weight="1.0" />
</bb_hbond_tors_fadun_cst>

The pre-selection mover

As an optional alternative to restricting oneself to backbone-only score terms, GeneralizedKIC permits the user to specify a pre-selection mover (pre_selection_mover="&string") defined prior to the GeneralizedKIC block. This permits side-chain moves, such as repacking or side-chain minimization, prior to scoring with the full energy function. Note that this mover will be applied to all candidate solutions passing filters, which makes this option potentially very computationally expensive!

If a pre-selection mover is specified, the mover may alter geometry outside of the loop to be closed, or may alter the FoldTree, in which case these alterations will carry through to the final pose returned by GeneralizedKIC. TaskOperations and MoveMaps may be used appropriately in the definition of the mover in question to restrict its effects to the loop on which GeneralizedKIC is operating if the user does not wish to alter geometry outside of this loop. If one wishes to apply more than one pre-selection mover (for example, a PackRotamersMover followed by a sidechain-only MinMover, to repack and minimize side-chains), they may be combined in a ParsedProtocol mover.