Score Terms#

TMol score functions combine weighted tmol.score.ScoreType values. A tmol.score.ScoreFunction activates the energy term implementation associated with each non-zero weight, then renders a whole-pose, block-pair, or rotamer scoring module for a particular tmol.pose.PoseStack.

Score-type map#

Score types	Energy-term class	Quantity represented
`fa_ljatr`, `fa_ljrep`, `fa_lk`	`LJLKEnergyTerm`	Attractive and repulsive Lennard-Jones interactions and isotropic Lazaridis-Karplus solvation.
`fa_elec`	`ElecEnergyTerm`	Distance-dependent full-atom electrostatics.
`hbond`	`HBondEnergyTerm`	Orientation-dependent hydrogen-bond interactions.
`cart_lengths`, `cart_angles`, `cart_torsions`, `cart_impropers`, `cart_hxltorsions`	`CartBondedEnergyTerm`	Bonded geometry used by Cartesian scoring and minimization.
`constraint`	`ConstraintEnergyTerm`	Harmonic distance, bounded distance, harmonic coordinate, and circular-harmonic torsion constraints attached through `tmol.pose.ConstraintSet`.
`disulfide`	`DisulfideEnergyTerm`	Geometry of disulfide-linked cysteine pairs.
`omega`, `rama`	`BackboneTorsionEnergyTerm`	Backbone-dependent peptide omega and Ramachandran torsion preferences.
`dunbrack_rot`, `dunbrack_rotdev`, `dunbrack_semirot`	`DunbrackEnergyTerm`	Backbone-dependent amino-acid rotamer probabilities and deviations.
`lk_ball_iso`, `lk_ball`, `lk_bridge`, `lk_bridge_uncpl`	`LKBallEnergyTerm`	Directional water-mediated solvation and bridging terms.
`ref`	`RefEnergyTerm`	Per-residue-type reference energies.
`gen_torsions`	`GenBondedEnergyTerm`	Generic bonded torsions, including ligand torsional parameters.
`na_torsion`, `na_torsion_well`	`NaTorsionEnergyTerm`	DNA/RNA backbone, glycosidic-chi, sugar, coupling, and rotamer-well preferences with polymer-specific parameters.

ScoreType.n_score_types is a terminal size sentinel used for weight-vector allocation; it is not a score component.

Non-bonded and solvation terms#

class tmol.score.ljlk.LJLKEnergyTerm(param_db: ParameterDatabase, device: device)[source]#: Bases: AtomTypeDependentTerm, BondDependentTerm

class tmol.score.elec.ElecEnergyTerm(param_db: ParameterDatabase, device: device)[source]#: Bases: AtomTypeDependentTerm, BondDependentTerm

class tmol.score.hbond.HBondEnergyTerm(param_db: ParameterDatabase, device: device)[source]#: Bases: AtomTypeDependentTerm, HBondDependentTerm

class tmol.score.lk_ball.LKBallEnergyTerm(param_db: ParameterDatabase, device: device)[source]#: Bases: AtomTypeDependentTerm, HBondDependentTerm

Bonded, torsional, and reference terms#

class tmol.score.cartbonded.CartBondedEnergyTerm(param_db: ParameterDatabase, device: device)[source]#: Bases: AtomTypeDependentTerm

class tmol.score.genbonded.GenBondedEnergyTerm(param_db: ParameterDatabase, device: device)[source]#: Bases: AtomTypeDependentTerm

class tmol.score.disulfide.DisulfideEnergyTerm(param_db: ParameterDatabase, device: device)[source]#: Bases: EnergyTerm

class tmol.score.backbone_torsion.BackboneTorsionEnergyTerm(param_db: ParameterDatabase, device: device)[source]#: Bases: EnergyTerm

class tmol.score.dunbrack.DunbrackEnergyTerm(param_db: ParameterDatabase, device: device)[source]#: Bases: EnergyTerm

class tmol.score.ref.RefEnergyTerm(param_db: ParameterDatabase, device: device)[source]#: Bases: EnergyTerm

Nucleic-acid term#

class tmol.score.na_torsion.NaTorsionEnergyTerm(param_db: ParameterDatabase, device: device)[source]#

Bases: EnergyTerm

Score nucleic-acid backbone, sugar-pucker, and glycosidic torsions.

Nucleic-acid torsion parameter resolution and scoring.

class tmol.score.na_torsion.NaTorsionParams(backbone_means: Tensor[2, 6, 3], backbone_n_bins: Tensor[2, 6], backbone_sdev: Tensor[2, 6], sugar_means: Tensor[2, 10, 4], chi_means: Tensor[8, 10], well_pucker: Tensor[2, 10], well_alpha_gamma: Tensor[2, 3, 3], well_bibii_pucker: Tensor[2, 2, 2], well_alphanext_bibii: Tensor[2, 3, 2], well_chi_syn: Tensor[2, 10, 8], is_north: Tensor, sdev_sugar: Tensor, sdev_chi: Tensor, weight_bb: Tensor, weight_chi: Tensor, weight_sugar: Tensor, pucker_temperature: float, bin_blend_sdev: float)[source]#

Bases: ValidateAttrs

Device-resident nucleic-acid torsion distributions and well energies.

tmol.score.na_torsion.bi_bii_weight(epsilon, zeta)[source]#

Weight on BI, from sin(epsilon - zeta).

exp(-s*del) / (exp(-s*del) + exp(s.del)) with s = 20

tmol.score.na_torsion.block_type_params(block_type, element_for_atom_type)[source]#

Per-block-type indices this term needs, shared by scoring and packing.

base is -1 for anything this term does not handle, including a nucleotide whose sugar or glycosidic torsions cannot be resolved.

tmol.score.na_torsion.eval_na_torsion_for_pose(rot_coords, _rot_coord_offset, _pose_ind_for_atom, _first_rot_for_block, _first_rot_block_type, _block_ind_for_rot, _pose_ind_for_rot, block_type_ind_for_rot, _n_rots_for_pose, _rot_offset_for_pose, _n_rots_for_block, _rot_offset_for_block, _max_n_rots_per_pose, has_na, pose_base, pose_is_na, pose_torsion_indices, pose_torsion_ok, pose_ring_indices, pose_ring_ok, pose_prev, bt_base, bt_uaids, bt_ring, bt_down, atom_downstream_of_conn, block_coord_offset, inter_residue_connections, backbone_means, backbone_sdev, sugar_means, chi_means, sdev_sugar, sdev_chi, well_pucker, well_alpha_gamma, well_bibii_pucker, well_alphanext_bibii, well_chi_syn, is_north, pucker_temperature, bin_blend_sdev, weight_bb, weight_chi, weight_sugar, output_block_pair_energies: bool)[source]#: Evaluate nucleic-acid torsion energies for packed rotamer coordinates.

tmol.score.na_torsion.na_torsion_subterms(rot_coords, block_type_ind_for_rot, bt_base, bt_uaids, bt_ring, bt_down, atom_downstream_of_conn, block_coord_offset, inter_residue_connections, backbone_means, backbone_sdev, sugar_means, chi_means, sdev_sugar, sdev_chi, well_pucker, well_alpha_gamma, well_bibii_pucker, well_alphanext_bibii, well_chi_syn, is_north, pucker_temperature, bin_blend_sdev)[source]#

Per-block (bb, chi, sugar, well) energies, the mask, and the base index.

The base index carries the polymer, which the caller needs in order to pick the per-polymer subterm weights.

tmol.score.na_torsion.polymer_index(base)[source]#: 0 for DNA, 1 for RNA. Non-nucleotides (base < 0) fall to 0 and are masked.

tmol.score.na_torsion.pucker_weights(ring, temperature)[source]#

Soft distribution over the 10 sugar pucker states.

ring: (…, 5, 3) ring coordinates in cyclic order, heteroatom last.

Walks the 5 cyclic rotations of the ring. Within each, dot measures how planar the leading four atoms are and exxo which face the fifth sits on. Softmax over -dot picks the apex; a sigmoid on exxo splits endo from exo.

tmol.score.na_torsion.sugar_ring_atoms(block_type, element_for_atom_type)[source]#

Ordered sugar ring, derived from the nu torsions rather than named.

nu0 and nu1 each span four consecutive ring atoms offset by one, so together they give the whole cycle in order. The pucker slot arithmetic is defined relative to a cycle ending on the ring heteroatom, so rotate to put it last.

tmol.score.na_torsion.syn_weight(chi, width=5.0)[source]#: Soft version of Rosetta’s chi in (20, 100) syn window.

The public parameter aliases and fitted-model constants are also available for lower-level analysis and extension work.

BACKBONE_TORSIONS
BASES
BASE_FOR_NAME3
CHI
CHI_TORSION

DELTA
NORTH_PUCKERS
NaTorsionParams
POLYMERS
REQUIRED_TORSIONS

SUGAR_TORSIONS
SYN_MEAN
SYN_RANGE
TORSION_NAMES
eval_na_torsion_for_pose

Constraints#

class tmol.score.constraint.ConstraintEnergyTerm(param_db: ParameterDatabase, device: device)[source]#

Bases: EnergyTerm

Score coordinate and geometric constraints attached to a pose stack.

Coordinate, distance, angle, and dihedral constraints.

class tmol.score.constraint.MCAtomIndices(max_n_mainchain_atoms: int, n_mainchain_atoms: Tensor, mainchain_atoms: Tensor[slice(None, None, None), slice(None, None, None)], is_real_mainchain_atom: Tensor[slice(None, None, None), slice(None, None, None)])[source]#

Bases: object

Padded main-chain atom indices for every packed block type.

tmol.score.constraint.constrain_all_ca(pose_stack: PoseStack) → PoseStack[source]#: Add coordinate constraints at every available protein CA atom.

tmol.score.constraint.create_mainchain_coordinate_constraints(pose_stack: PoseStack) → PoseStack[source]#: Add coordinate constraints for every declared polymer main-chain atom.

`MCAtomIndices`	Padded main-chain atom indices for every packed block type.
`constrain_all_ca`	Add coordinate constraints at every available protein CA atom.
`create_mainchain_coordinate_constraints`	Add coordinate constraints for every declared polymer main-chain atom.

Score-term construction#

The creator classes below are the public extension surface used by tmol.score.terms.ScoreTermFactory. Most applications should configure a tmol.score.ScoreFunction rather than instantiate creators directly.

Factories for constructing standard score terms.

class tmol.score.terms.BackboneTorsionTermCreator[source]#

Bases: TermCreator

Create the protein backbone torsion energy term.

class tmol.score.terms.CartBondedTermCreator[source]#

Bases: TermCreator

Create the Cartesian bonded-geometry energy term.

class tmol.score.terms.ConstraintTermCreator[source]#

Bases: TermCreator

Create the pose-constraint energy term.

class tmol.score.terms.DisulfideTermCreator[source]#

Bases: TermCreator

Create the disulfide-geometry energy term.

class tmol.score.terms.DunbrackTermCreator[source]#

Bases: TermCreator

Create the Dunbrack rotamer energy term.

class tmol.score.terms.ElecTermCreator[source]#

Bases: TermCreator

Create the electrostatic energy term.

class tmol.score.terms.GenBondedTermCreator[source]#

Bases: TermCreator

Create the general bonded-torsion energy term.

class tmol.score.terms.HBondTermCreator[source]#

Bases: TermCreator

Create the hydrogen-bond energy term.

class tmol.score.terms.LJLKTermCreator[source]#

Bases: TermCreator

Create the Lennard-Jones and Lazaridis-Karplus energy term.

class tmol.score.terms.LKBallTermCreator[source]#

Bases: TermCreator

Create the orientation-dependent LK-ball solvation energy term.

class tmol.score.terms.NaTorsionTermCreator[source]#

Bases: TermCreator

Create the nucleic-acid torsion energy term.

class tmol.score.terms.RefTermCreator[source]#

Bases: TermCreator

Create the residue reference-energy term.

class tmol.score.terms.ScoreTermFactory[source]#

Bases: object

Factory for the creation of EnergyTerms

This class uses import-time factory registration to discover the set of TermCreators that live in the same directory as it. To register a new TermCreator, simply put the term creator in this directory.

class tmol.score.terms.TermCreator[source]#

Bases: object

Base class for registering score terms with the ScoreTermFactory.

To add a new term,

add one or more new entries to the tmol.score._score_types enumeration

derive a new subclass of TermCreator and put it in this directory (the term itself should be implemented in a different directory)

the new TermCreator subclass needs to define two methods, create_term and score_types

create_term should instantiate the term

score_types should return a list of the elements of the score_types enumeration that the term implements in the order that the term will report them

tmol.score.terms.score_term_creator(cls)[source]#: Register a term-creator class with the global score-term factory.