Distance geometry-related classes and functions. More...

Data Structures
struct	Configuration
	A configuration object for distance geometry runs with sane defaults. More...

struct	MoleculeDGInformation
	Intermediate conformational data about a Molecule given by a spatial model. More...

class	DistanceBoundsMatrix

struct	ChiralConstraint
	Data struct representing a chiral constraint. More...

struct	DihedralConstraint
	Data struct representing a dihedral constraint. More...

class	EigenRefinementProblem
	Eigen-based refinement error function. More...

struct	RefinementTraits
	This is for when you have a fully qualified Refinement typename but want to get the template arguments back. More...

class	ExplicitBoundsGraph
	BGL wrapper to help with distance bounds smoothing. More...

class	ImplicitBoundsGraph
	Simulates a graph from which triangle inequality bounds can be calculated by shortest-paths. More...

class	MetricMatrix

class	SpatialModel
	Class performing spatial modeling of molecules. More...

struct	ValueBounds
	Data class for bounded values. More...

Enumerations
enum	Partiality { Partiality::FourAtom, Partiality::TenPercent, Partiality::All }
	Limit triangle inequality bounds smoothing to a subset of all atoms. More...

Functions
MoleculeDGInformation	gatherDGInformation (const Molecule &molecule, const Configuration &configuration)
	Collects intermediate conformational data about a Molecule using a spatial model. More...

Result< AngstromPositions >	refine (Eigen::MatrixXd embeddedPositions, const DistanceBoundsMatrix &distanceBounds, const Configuration &configuration, const std::shared_ptr< MoleculeDGInformation > &DgDataPtr)
	Distance Geometry refinement.

Result< AngstromPositions >	generateConformer (const Molecule &molecule, const Configuration &configuration, std::shared_ptr< MoleculeDGInformation > &DgDataPtr, bool regenerateDGDataEachStep, Random::Engine &engine)

std::vector< Result < AngstromPositions >>	run (const Molecule &molecule, unsigned numConformers, const Configuration &configuration, boost::optional< unsigned > seedOption)
	Main and parallel implementation of Distance Geometry. Generates an ensemble of 3D structures of a given Molecule. More...

template<class RefinementType , typename PositionType >
bool	finalStructureAcceptable (const RefinementType &refinement, const DistanceBoundsMatrix &bounds, const PositionType &positions)
	Decides whether the final structure from a refinement is acceptable. More...

unsigned	tetrangleSmooth (Eigen::Ref< Eigen::MatrixXd > bounds)
	Smoothes the bounds matrix using tetrangle inequalities. More...

Detailed Description

Distance geometry-related classes and functions.

Enumeration Type Documentation

enum Scine::Molassembler::DistanceGeometry::Partiality

strong

Limit triangle inequality bounds smoothing to a subset of all atoms.

Usually, after choosing a single conformer's atom-pairwise distances from between the distance bounds that are generated from the spatial model, all other distance bounds are re-smoothed using the triangle inequality. However, the reduction in distance bounds slack all over the matrix diminishes with each successive chosen distance, yielding only diminishing returns.

With this option, you can choose to stop re-smoothing the entire matrix after a limited number of one-to-all distance choices. This has the following tradeoff effects:

Reduces computational effort in generating a distance matrix, speeding up conformer generation
Worsens the quality of initial embedded coordinates prior to refinement, requiring more refinement steps

Enumerator

FourAtom

Perform smoothing for four one-to-all distance choices.

In principle, if the distances from four atoms to all others are known, the overall conformation is fully determined. This is unfortunately not realized in practice in the DG procedure.

This yields the most speedup in distance matrix generation, but also worsens the initial embedded coordinates the most.

TenPercent

Perform smoothing for ten percent of one-to-all distance choices.

Note: Still performs bounds smoothing for at least four one-to-all distance choices if the number of atoms is less than 40.

All

Perform smoothing after all distance choices.

This yields the slowest distance matrix generation, but also the best initial embedded coordinates.

Function Documentation

template<class RefinementType , typename PositionType >

bool Scine::Molassembler::DistanceGeometry::finalStructureAcceptable	(	const RefinementType &	refinement,
		const DistanceBoundsMatrix &	bounds,
		const PositionType &	positions
	)

Decides whether the final structure from a refinement is acceptable.

A final structure is acceptable if

All distance bounds are within 0.5 of either the lower or upper boundary
All chiral constraints are within 0.5 of either the lower or upper boundary

Complexity \(\Theta(N^2)\) due to atom-pairwise distance bounds

Parameters

refinement	The refinement functor
bounds	The distance bounds
positions	The final positions from a refinement

Returns: Whether the final structure is acceptable

MoleculeDGInformation Scine::Molassembler::DistanceGeometry::gatherDGInformation	(	const Molecule &	molecule,
		const Configuration &	configuration
	)

Collects intermediate conformational data about a Molecule using a spatial model.

Complexity At least \(O(P_2 + P_3 + P_4)\) where \(P_i\) is the number of distinct paths of length \(i\) in the graph. That should scale at least linearly in the number of vertices.

std::vector< Result<AngstromPositions>> Scine::Molassembler::DistanceGeometry::run	(	const Molecule &	molecule,
		unsigned	numConformers,
		const Configuration &	configuration,
		boost::optional< unsigned >	seedOption
	)

Main and parallel implementation of Distance Geometry. Generates an ensemble of 3D structures of a given Molecule.

Complexity Roughly \(O(C \cdot N^3)\) where \(C\) is the number of conformers and \(N\) is the number of atoms in molecule

See Also: generateEnsemble

unsigned Scine::Molassembler::DistanceGeometry::tetrangleSmooth ( Eigen::Ref< Eigen::MatrixXd > bounds )

Smoothes the bounds matrix using tetrangle inequalities.

Precondition: Assumes bounds has already been triangle inequality smoothed

Parameters

bounds A square matrix with zeros on the diagonal and B(j, i) <= B(i, j) for all i < j (lower bounds on strictly lower triangle, upper bounds on strictly upper triangle)

Complexity \(\Theta(N^4)\)

Warning: This doesn't work and I don't know why.

Returns: Tetrangle smoothed bounds matrix

Data Structures

Enumerations

Functions

Detailed Description

Enumeration Type Documentation

Function Documentation