Contains general bit-comparison and similarity operations. More...

#include <RDGeneral/export.h>
#include "BitVects.h"
#include <string>

Functions
template<typename T >
double	SimilarityWrapper (const T &bv1, const T &bv2, double(*metric)(const T &, const T &), bool returnDistance=false)

template<typename T >
double	SimilarityWrapper (const T &bv1, const T &bv2, double a, double b, double(*metric)(const T &, const T &, double, double), bool returnDistance=false)
	This is an overloaded member function, provided for convenience. It differs from the above function only in what argument(s) it accepts.

RDKIT_DATASTRUCTS_EXPORT bool	AllProbeBitsMatch (const char probe, const char ref)

RDKIT_DATASTRUCTS_EXPORT bool	AllProbeBitsMatch (const std::string &probe, const std::string &ref)

RDKIT_DATASTRUCTS_EXPORT bool	AllProbeBitsMatch (const ExplicitBitVect &probe, const ExplicitBitVect &ref)

template<typename T1 >
RDKIT_DATASTRUCTS_EXPORT bool	AllProbeBitsMatch (const T1 &probe, const std::string &pkl)

template<typename T1 >
RDKIT_DATASTRUCTS_EXPORT bool	AllProbeBitsMatch (const T1 &probe, const T1 &ref)

template<typename T1 , typename T2 >
RDKIT_DATASTRUCTS_EXPORT int	NumOnBitsInCommon (const T1 &bv1, const T2 &bv2)
	returns the number of on bits in common between two bit vectors

RDKIT_DATASTRUCTS_EXPORT int	NumOnBitsInCommon (const ExplicitBitVect &bv1, const ExplicitBitVect &bv2)

template<typename T1 , typename T2 >
RDKIT_DATASTRUCTS_EXPORT double	TanimotoSimilarity (const T1 &bv1, const T2 &bv2)
	returns the Tanimoto similarity between two bit vects

template<typename T1 , typename T2 >
RDKIT_DATASTRUCTS_EXPORT double	CosineSimilarity (const T1 &bv1, const T2 &bv2)
	returns the Cosine similarity between two bit vects

template<typename T1 , typename T2 >
RDKIT_DATASTRUCTS_EXPORT double	KulczynskiSimilarity (const T1 &bv1, const T2 &bv2)
	returns the Kulczynski similarity between two bit vects

template<typename T1 , typename T2 >
RDKIT_DATASTRUCTS_EXPORT double	DiceSimilarity (const T1 &bv1, const T2 &bv2)
	returns the Dice similarity between two bit vects

template<typename T1 , typename T2 >
RDKIT_DATASTRUCTS_EXPORT double	TverskySimilarity (const T1 &bv1, const T2 &bv2, double a, double b)
	returns the Tversky similarity between two bit vects

template<typename T1 , typename T2 >
RDKIT_DATASTRUCTS_EXPORT double	SokalSimilarity (const T1 &bv1, const T2 &bv2)
	returns the Sokal similarity between two bit vects

template<typename T1 , typename T2 >
RDKIT_DATASTRUCTS_EXPORT double	McConnaugheySimilarity (const T1 &bv1, const T2 &bv2)
	returns the McConnaughey similarity between two bit vects

template<typename T1 , typename T2 >
RDKIT_DATASTRUCTS_EXPORT double	AsymmetricSimilarity (const T1 &bv1, const T2 &bv2)
	returns the Asymmetric similarity between two bit vects

template<typename T1 , typename T2 >
RDKIT_DATASTRUCTS_EXPORT double	BraunBlanquetSimilarity (const T1 &bv1, const T2 &bv2)
	returns the Braun-Blanquet similarity between two bit vects

template<typename T1 , typename T2 >
RDKIT_DATASTRUCTS_EXPORT double	RusselSimilarity (const T1 &bv1, const T2 &bv2)
	returns the Russel similarity between two bit vects

template<typename T1 , typename T2 >
RDKIT_DATASTRUCTS_EXPORT double	RogotGoldbergSimilarity (const T1 &bv1, const T2 &bv2)
	returns the Rogot-Goldberg similarity between two bit vects

template<typename T1 , typename T2 >
RDKIT_DATASTRUCTS_EXPORT double	OnBitSimilarity (const T1 &bv1, const T2 &bv2)
	returns the on bit similarity between two bit vects

template<typename T1 , typename T2 >
RDKIT_DATASTRUCTS_EXPORT int	NumBitsInCommon (const T1 &bv1, const T2 &bv2)
	returns the number of common bits (on and off) between two bit vects

RDKIT_DATASTRUCTS_EXPORT int	NumBitsInCommon (const ExplicitBitVect &bv1, const ExplicitBitVect &bv2)

template<typename T1 , typename T2 >
RDKIT_DATASTRUCTS_EXPORT double	AllBitSimilarity (const T1 &bv1, const T2 &bv2)

template<typename T1 , typename T2 >
RDKIT_DATASTRUCTS_EXPORT IntVect	OnBitsInCommon (const T1 &bv1, const T2 &bv2)

template<typename T1 , typename T2 >
RDKIT_DATASTRUCTS_EXPORT IntVect	OffBitsInCommon (const T1 &bv1, const T2 &bv2)

template<typename T1 , typename T2 >
RDKIT_DATASTRUCTS_EXPORT DoubleVect	OnBitProjSimilarity (const T1 &bv1, const T2 &bv2)
	returns the on-bit projected similarities between two bit vects

template<typename T1 , typename T2 >
RDKIT_DATASTRUCTS_EXPORT DoubleVect	OffBitProjSimilarity (const T1 &bv1, const T2 &bv2)
	returns the on-bit projected similarities between two bit vects

template<typename T1 >
RDKIT_DATASTRUCTS_EXPORT T1 *	FoldFingerprint (const T1 &bv1, unsigned int factor=2)
	folds a bit vector `factor` times and returns the result

template<typename T1 >
RDKIT_DATASTRUCTS_EXPORT std::string	BitVectToText (const T1 &bv1)
	returns a text representation of a bit vector (a string of 0s and 1s)

template<typename T1 >
RDKIT_DATASTRUCTS_EXPORT std::string	BitVectToFPSText (const T1 &bv1)

template<typename T1 >
RDKIT_DATASTRUCTS_EXPORT std::string	BitVectToBinaryText (const T1 &bv1)
	returns a binary string representation of a bit vector (an array of bytes)

template<typename T1 >
RDKIT_DATASTRUCTS_EXPORT void	UpdateBitVectFromFPSText (T1 &bv1, const std::string &fps)
	updates a bit vector from Andrew Dalke's FPS format

template<typename T1 >
RDKIT_DATASTRUCTS_EXPORT void	UpdateBitVectFromBinaryText (T1 &bv1, const std::string &fps)

RDKIT_DATASTRUCTS_EXPORT unsigned int	CalcBitmapPopcount (const unsigned char *bv1, unsigned int nBytes)

RDKIT_DATASTRUCTS_EXPORT double	CalcBitmapTanimoto (const unsigned char bv1, const unsigned char bv2, unsigned int nBytes)

RDKIT_DATASTRUCTS_EXPORT double	CalcBitmapDice (const unsigned char bv1, const unsigned char bv2, unsigned int nBytes)

RDKIT_DATASTRUCTS_EXPORT double	CalcBitmapTversky (const unsigned char bv1, const unsigned char bv2, unsigned int nBytes, double ca, double cb)

RDKIT_DATASTRUCTS_EXPORT bool	CalcBitmapAllProbeBitsMatch (const unsigned char probe, const unsigned char ref, unsigned int nBytes)

Detailed Description

Contains general bit-comparison and similarity operations.

The notation used to document the similarity metrics is:

V1_n: number of bits in vector 1
V1_o: number of on bits in vector 1
(V1&V2)_o: number of on bits in the intersection of vectors 1 and 2

Definition in file BitOps.h.

Function Documentation

◆ AllBitSimilarity()

template<typename T1 , typename T2 >

RDKIT_DATASTRUCTS_EXPORT double AllBitSimilarity	(	const T1 &	bv1,
		const T2 &	bv2
	)

returns the common-bit similarity (on and off) between two bit vects This is also called Manhattan similarity.

Returns: [bv1_n - (bv1^bv2)_o] / bv1_n

◆ AllProbeBitsMatch() [1/5]

RDKIT_DATASTRUCTS_EXPORT bool AllProbeBitsMatch	(	const char *	probe,
		const char *	ref
	)

Referenced by RDKit::FPHolderBase::passesFilter().

◆ AllProbeBitsMatch() [2/5]

RDKIT_DATASTRUCTS_EXPORT bool AllProbeBitsMatch	(	const ExplicitBitVect &	probe,
		const ExplicitBitVect &	ref
	)

◆ AllProbeBitsMatch() [3/5]

RDKIT_DATASTRUCTS_EXPORT bool AllProbeBitsMatch	(	const std::string &	probe,
		const std::string &	ref
	)

◆ AllProbeBitsMatch() [4/5]

template<typename T1 >

RDKIT_DATASTRUCTS_EXPORT bool AllProbeBitsMatch	(	const T1 &	probe,
		const std::string &	pkl
	)

◆ AllProbeBitsMatch() [5/5]

template<typename T1 >

RDKIT_DATASTRUCTS_EXPORT bool AllProbeBitsMatch	(	const T1 &	probe,
		const T1 &	ref
	)

◆ AsymmetricSimilarity()

template<typename T1 , typename T2 >

RDKIT_DATASTRUCTS_EXPORT double AsymmetricSimilarity	(	const T1 &	bv1,
		const T2 &	bv2
	)

returns the Asymmetric similarity between two bit vects

Returns: (bv1&bv2)_o / min(bv1_o,bv2_o)

◆ BitVectToBinaryText()

template<typename T1 >

RDKIT_DATASTRUCTS_EXPORT std::string BitVectToBinaryText ( const T1 & bv1 )

returns a binary string representation of a bit vector (an array of bytes)

Parameters

bv1	the vector to use

Returns: an std::string

◆ BitVectToFPSText()

template<typename T1 >

RDKIT_DATASTRUCTS_EXPORT std::string BitVectToFPSText ( const T1 & bv1 )

returns a hex representation of a bit vector compatible with Andrew Dalke's FPS format

Parameters

bv1	the vector to use

Returns: an std::string

◆ BitVectToText()

template<typename T1 >

RDKIT_DATASTRUCTS_EXPORT std::string BitVectToText ( const T1 & bv1 )

returns a text representation of a bit vector (a string of 0s and 1s)

Parameters

bv1	the vector to use

Returns: an std::string

◆ BraunBlanquetSimilarity()

template<typename T1 , typename T2 >

RDKIT_DATASTRUCTS_EXPORT double BraunBlanquetSimilarity	(	const T1 &	bv1,
		const T2 &	bv2
	)

returns the Braun-Blanquet similarity between two bit vects

Returns: (bv1&bv2)_o / max(bv1_o,bv2_o)

◆ CalcBitmapAllProbeBitsMatch()

RDKIT_DATASTRUCTS_EXPORT bool CalcBitmapAllProbeBitsMatch	(	const unsigned char *	probe,
		const unsigned char *	ref,
		unsigned int	nBytes
	)

◆ CalcBitmapDice()

RDKIT_DATASTRUCTS_EXPORT double CalcBitmapDice	(	const unsigned char *	bv1,
		const unsigned char *	bv2,
		unsigned int	nBytes
	)

◆ CalcBitmapPopcount()

RDKIT_DATASTRUCTS_EXPORT unsigned int CalcBitmapPopcount	(	const unsigned char *	bv1,
		unsigned int	nBytes
	)

◆ CalcBitmapTanimoto()

RDKIT_DATASTRUCTS_EXPORT double CalcBitmapTanimoto	(	const unsigned char *	bv1,
		const unsigned char *	bv2,
		unsigned int	nBytes
	)

◆ CalcBitmapTversky()

RDKIT_DATASTRUCTS_EXPORT double CalcBitmapTversky	(	const unsigned char *	bv1,
		const unsigned char *	bv2,
		unsigned int	nBytes,
		double	ca,
		double	cb
	)

◆ CosineSimilarity()

template<typename T1 , typename T2 >

RDKIT_DATASTRUCTS_EXPORT double CosineSimilarity	(	const T1 &	bv1,
		const T2 &	bv2
	)

returns the Cosine similarity between two bit vects

Returns: (bv1&bv2)_o / sqrt(bv1_o + bv2_o)

◆ DiceSimilarity()

template<typename T1 , typename T2 >

RDKIT_DATASTRUCTS_EXPORT double DiceSimilarity	(	const T1 &	bv1,
		const T2 &	bv2
	)

returns the Dice similarity between two bit vects

Returns: 2*(bv1&bv2)_o / [bv1_o + bv2_o]

◆ FoldFingerprint()

template<typename T1 >

RDKIT_DATASTRUCTS_EXPORT T1 * FoldFingerprint	(	const T1 &	bv1,
		unsigned int	factor = `2`
	)

folds a bit vector factor times and returns the result

Parameters

bv1	the vector to be folded
factor	(optional) the number of times to fold it

Returns: a pointer to the folded fingerprint, which is bv1_n/factor long.

Note: The caller is responsible for deleteing the result.

Referenced by SimilarityWrapper(), and SimilarityWrapper().

◆ KulczynskiSimilarity()

template<typename T1 , typename T2 >

RDKIT_DATASTRUCTS_EXPORT double KulczynskiSimilarity	(	const T1 &	bv1,
		const T2 &	bv2
	)

returns the Kulczynski similarity between two bit vects

Returns: (bv1&bv2)_o * [bv1_o + bv2_o] / [2 * bv1_o * bv2_o]

◆ McConnaugheySimilarity()

template<typename T1 , typename T2 >

RDKIT_DATASTRUCTS_EXPORT double McConnaugheySimilarity	(	const T1 &	bv1,
		const T2 &	bv2
	)

returns the McConnaughey similarity between two bit vects

Returns: [(bv1&bv2)_o * (bv1_o + bv2_o) - (bv1_o * bv2_o)] / (bv1_o * bv2_o)

◆ NumBitsInCommon() [1/2]

RDKIT_DATASTRUCTS_EXPORT int NumBitsInCommon	(	const ExplicitBitVect &	bv1,
		const ExplicitBitVect &	bv2
	)

◆ NumBitsInCommon() [2/2]

template<typename T1 , typename T2 >

RDKIT_DATASTRUCTS_EXPORT int NumBitsInCommon	(	const T1 &	bv1,
		const T2 &	bv2
	)

returns the number of common bits (on and off) between two bit vects

Returns: bv1_n - (bv1^bv2)_o

◆ NumOnBitsInCommon() [1/2]

RDKIT_DATASTRUCTS_EXPORT int NumOnBitsInCommon	(	const ExplicitBitVect &	bv1,
		const ExplicitBitVect &	bv2
	)

◆ NumOnBitsInCommon() [2/2]

template<typename T1 , typename T2 >

RDKIT_DATASTRUCTS_EXPORT int NumOnBitsInCommon	(	const T1 &	bv1,
		const T2 &	bv2
	)

returns the number of on bits in common between two bit vectors

Returns: (bv1&bv2)_o

◆ OffBitProjSimilarity()

template<typename T1 , typename T2 >

RDKIT_DATASTRUCTS_EXPORT DoubleVect OffBitProjSimilarity	(	const T1 &	bv1,
		const T2 &	bv2
	)

returns the on-bit projected similarities between two bit vects

Returns

two values, as a DoubleVect:

[bv1_n - (bv1|bv2)_o] / [bv1_n - bv1_o]
[bv2_n - (bv1|bv2)_o] / [bv2_n - bv2_o]

Note: bv1_n = bv2_n

◆ OffBitsInCommon()

template<typename T1 , typename T2 >

RDKIT_DATASTRUCTS_EXPORT IntVect OffBitsInCommon	(	const T1 &	bv1,
		const T2 &	bv2
	)

returns an IntVect with indices of all off bits in common between two bit vects

◆ OnBitProjSimilarity()

template<typename T1 , typename T2 >

RDKIT_DATASTRUCTS_EXPORT DoubleVect OnBitProjSimilarity	(	const T1 &	bv1,
		const T2 &	bv2
	)

returns the on-bit projected similarities between two bit vects

Returns

two values, as a DoubleVect:

(bv1&bv2)_o / bv1_o
(bv1&bv2)_o / bv2_o

◆ OnBitSimilarity()

template<typename T1 , typename T2 >

RDKIT_DATASTRUCTS_EXPORT double OnBitSimilarity	(	const T1 &	bv1,
		const T2 &	bv2
	)

returns the on bit similarity between two bit vects

Returns: (bv1&bv2)_o / (bv1|bv2)_o

◆ OnBitsInCommon()

template<typename T1 , typename T2 >

RDKIT_DATASTRUCTS_EXPORT IntVect OnBitsInCommon	(	const T1 &	bv1,
		const T2 &	bv2
	)

returns an IntVect with indices of all on bits in common between two bit vects

◆ RogotGoldbergSimilarity()

template<typename T1 , typename T2 >

RDKIT_DATASTRUCTS_EXPORT double RogotGoldbergSimilarity	(	const T1 &	bv1,
		const T2 &	bv2
	)

returns the Rogot-Goldberg similarity between two bit vects

Returns

(bv1&bv2)_o / (bv1_o + bv2_o)
(bv1_n - bv1_o - bv2_o + (bv1&bv2)_o) / (2*bv1_n - bv1_o - bv2_o)

◆ RusselSimilarity()

template<typename T1 , typename T2 >

RDKIT_DATASTRUCTS_EXPORT double RusselSimilarity	(	const T1 &	bv1,
		const T2 &	bv2
	)

returns the Russel similarity between two bit vects

Returns: (bv1&bv2)_o / bv1_o

Note: that this operation is non-commutative: RusselSimilarity(bv1,bv2) != RusselSimilarity(bv2,bv1)

◆ SimilarityWrapper() [1/2]

template<typename T >

double SimilarityWrapper	(	const T &	bv1,
		const T &	bv2,
		double	a,
		double	b,
		double(*)(const T &, const T &, double, double)	metric,
		bool	returnDistance = `false`
	)

This is an overloaded member function, provided for convenience. It differs from the above function only in what argument(s) it accepts.

Definition at line 53 of file BitOps.h.

References FoldFingerprint().

◆ SimilarityWrapper() [2/2]

template<typename T >

double SimilarityWrapper	(	const T &	bv1,
		const T &	bv2,
		double(*)(const T &, const T &)	metric,
		bool	returnDistance = `false`
	)

general purpose wrapper for calculating the similarity between two bvs that may be of unequal size (will automatically fold as appropriate)

Definition at line 31 of file BitOps.h.

References FoldFingerprint().

Referenced by RDDataManip::TanimotoDistanceMetric(), and RDDataManip::TanimotoSimilarityMetric().

◆ SokalSimilarity()

template<typename T1 , typename T2 >

RDKIT_DATASTRUCTS_EXPORT double SokalSimilarity	(	const T1 &	bv1,
		const T2 &	bv2
	)

returns the Sokal similarity between two bit vects

Returns: (bv1&bv2)_o / [2*bv1_o + 2*bv2_o - 3*(bv1&bv2)_o]

◆ TanimotoSimilarity()

template<typename T1 , typename T2 >

RDKIT_DATASTRUCTS_EXPORT double TanimotoSimilarity	(	const T1 &	bv1,
		const T2 &	bv2
	)

returns the Tanimoto similarity between two bit vects

Returns: (bv1&bv2)_o / [bv1_o + bv2_o - (bv1&bv2)_o]

Referenced by RDDataManip::TanimotoDistanceMetric(), and RDDataManip::TanimotoSimilarityMetric().

◆ TverskySimilarity()

template<typename T1 , typename T2 >

RDKIT_DATASTRUCTS_EXPORT double TverskySimilarity	(	const T1 &	bv1,
		const T2 &	bv2,
		double	a,
		double	b
	)

returns the Tversky similarity between two bit vects

Returns: (bv1&bv2)_o / [a*bv1_o + b*bv2_o + (1 - a - b)*(bv1&bv2)_o]

Notes:

0 <= a,b <= 1

Tversky(a=1,b=1) = Tanimoto

Tversky(a=1/2,b=1/2) = Dice

◆ UpdateBitVectFromBinaryText()

template<typename T1 >

RDKIT_DATASTRUCTS_EXPORT void UpdateBitVectFromBinaryText	(	T1 &	bv1,
		const std::string &	fps
	)

updates a bit vector from a binary string representation of a bit vector (an array of bytes)

Parameters

bv1	the vector to use
fps	the binary string

◆ UpdateBitVectFromFPSText()

template<typename T1 >

RDKIT_DATASTRUCTS_EXPORT void UpdateBitVectFromFPSText	(	T1 &	bv1,
		const std::string &	fps
	)

updates a bit vector from Andrew Dalke's FPS format

Parameters

bv1	the vector to use
fps	the FPS hex string

Functions

Detailed Description

Function Documentation

◆ AllBitSimilarity()

◆ AllProbeBitsMatch() [1/5]

◆ AllProbeBitsMatch() [2/5]

◆ AllProbeBitsMatch() [3/5]

◆ AllProbeBitsMatch() [4/5]

◆ AllProbeBitsMatch() [5/5]

◆ AsymmetricSimilarity()

◆ BitVectToBinaryText()

◆ BitVectToFPSText()

◆ BitVectToText()

◆ BraunBlanquetSimilarity()

◆ CalcBitmapAllProbeBitsMatch()

◆ CalcBitmapDice()

◆ CalcBitmapPopcount()

◆ CalcBitmapTanimoto()

◆ CalcBitmapTversky()

◆ CosineSimilarity()

◆ DiceSimilarity()

◆ FoldFingerprint()

◆ KulczynskiSimilarity()

◆ McConnaugheySimilarity()

◆ NumBitsInCommon() [1/2]

◆ NumBitsInCommon() [2/2]

◆ NumOnBitsInCommon() [1/2]

◆ NumOnBitsInCommon() [2/2]

◆ OffBitProjSimilarity()

◆ OffBitsInCommon()

◆ OnBitProjSimilarity()

◆ OnBitSimilarity()

◆ OnBitsInCommon()

◆ RogotGoldbergSimilarity()

◆ RusselSimilarity()

◆ SimilarityWrapper() [1/2]

◆ SimilarityWrapper() [2/2]

◆ SokalSimilarity()

◆ TanimotoSimilarity()

◆ TverskySimilarity()

0 <= a,b <= 1

Tversky(a=1,b=1) = Tanimoto

Tversky(a=1/2,b=1/2) = Dice

◆ UpdateBitVectFromBinaryText()

◆ UpdateBitVectFromFPSText()