Package chemaxon.descriptors

Class Metrics

java.lang.Object

chemaxon.descriptors.Metrics

@PublicApi
public class Metrics
extends Object

The Metrics class implements dissimilarity metrics for MolecularDescriptors. For the sake of general availability of this dissimilarity calculations by any MolecularDescriptor derivatives, methods take float and int arrays as input parameters, thus they do not need to access to the internal representation of descriptors. This, however, may introduce some overhead in dissimilarity calculations, since MolecularDescriptor objects may need to transform their internal representation into an int or a float array. If this overhead is significant, then the required dissimilarity function has to be implemented as part of the descriptor class.

Since:

JChem 2.3

Constructor Summary

Constructors

Constructor	Description
Metrics()

Method Summary

Modifier and Type	Method	Description
static float	asymmetricEuclidean(float[] v1, float[] v2, double asymmetryFactor)	Calculates the asymmetric Euclidean distance.
static float	asymmetricEuclidean(int[] v1, int[] v2, double asymmetryFactor)	Calculates the asymmetric Euclidean distance.
static float	asymmetricNormalizedEuclidean(float[] v1, float[] v2, double asymmetryFactor)	Calculates the asymmetric normalized Euclidean distance.
static float	asymmetricNormalizedEuclidean(int[] v1, int[] v2, double asymmetryFactor)	Calculates the asymmetric normalized Euclidean distance.
static float	asymmetricTanimoto(float[] v1, float[] v2, float asymmetryFactor)	Calculates the asymmetric Tanimoto dissimilarity for floating point values.
static float	asymmetricTanimoto(int[] v1, int[] v2, float asymmetryFactor)	Calculates the asymmetric Tanimoto dissimilarity for floating point values.
static float	binaryAsymmetricEuclidean(int[] v1, int[] v2, float asymmetryFactor)	Calculates the asymmetric Euclidean distance of binary descriptors.
static float	binaryAsymmetricNormalizedEuclidean(int[] v1, int[] v2, float asymmetryFactor)	Calculates the asymmetric normalized Euclidean distance of binary descriptors.
static float	binaryAsymmetricNormalizedEuclidean(int[] v1, int oneCount1, int[] v2, int oneCount2, float asymmetryFactor)	Calculates the asymmetric normalized Euclidean distance of binary descriptors.
static float	binaryAsymmetricTanimoto(int[] v1, int[] v2, float asymmetryFactor)	Calculates asymmetric Tanimoto dissimilarity for binary descriptors.
static float	binaryAsymmetricTanimoto(int[] v1, int oneCount1, int[] v2, int oneCount2, float asymmetryFactor)	Calculates asymmetricTanimoto dissimilarity for binary descriptors.
static float	binaryEuclidean(int[] v1, int[] v2)	Calculates the Euclidean distance of two binary descriptors.
static float	binaryNormalizedEuclidean(int[] v1, int[] v2)	Calculates the normalized Euclidean distance of two binary descriptors.
static float	binaryNormalizedEuclidean(int[] v1, int oneCount1, int[] v2, int oneCount2)	Calculates the normalized Euclidean distance of two binary descriptors.
static float	binaryTanimoto(int[] v1, int[] v2)	Calculates Tanimoto dissimilarity for binary descriptors.
static float	binaryTanimoto(int[] v1, int oneCount1, int[] v2, int oneCount2)	Calculates Tanimoto dissimilarity for binary descriptors.
static float	binaryTversky(int[] v1, float wv1, int[] v2, float wv2)	Calculates Tversky dissimilarity for binary descriptors.
static float	binaryWeightedAsymmetricEuclidean(int[] v1, int[] v2, float[] weights, float asymmetryFactor)	Calculates the weighted asymmetric Euclidean distance of binary descriptors.
static float	binaryWeightedAsymmetricNormalizedEuclidean(int[] v1, int[] v2, float[] weights, float asymmetryFactor)	Calculates the weighted asymmetric normalized Euclidean distance of binary descriptors.
static float	binaryWeightedAsymmetricNormalizedEuclidean(int[] v1, int oneCount1, int[] v2, int oneCount2, float[] weights, float asymmetryFactor)	Calculates the weighted asymmetric normalized Euclidean distance of binary descriptors.
static float	binaryWeightedEuclidean(int[] v1, int[] v2, float[] weights)	Calculates the weighted Euclidean distance of binary descriptors.
static float	binaryWeightedNormalizedEuclidean(int[] v1, int[] v2, float[] weights)	Calculates the weighted normalized Euclidean distance of binary descriptors.
static float	binaryWeightedNormalizedEuclidean(int[] v1, int oneCount1, int[] v2, int oneCount2, int[] weights)	Calculates the weighted normalized Euclidean distance of binary descriptors.
static int	calcBitCount(int i)	Counts the number of bits set (1) in the integer parameter.
static int	calcBitCount(int[] v)	Counts the number of bits set (1) in a bit vector.
static float	euclidean(float[] v1, float[] v2)	Calculates the Euclidean distance.
static float	euclidean(int[] v1, int[] v2)	Calculates the Euclidean distance.
static float	normalizedEuclidean(float[] v1, float[] v2)	Calculates the normalized Euclidean distance.
static float	normalizedEuclidean(int[] v1, int[] v2)	Calculates the normalized Euclidean distance.
static float	scaledAsymmetricTanimoto(float[] v1, float[] v2, float[] hypothesis, float scaleFactor, float asymmetryFactor)	Calculates the scaled asymmetric Tanimoto dissimilarity for floating point values.
static float	scaledAsymmetricTanimoto(int[] v1, int[] v2, float[] hypothesis, float scaleFactor, float asymmetryFactor)	Calculates the scaled asymmetric Tanimoto dissimilarity for floating point values.
static float	scaledTanimoto(float[] v1, float[] v2, float[] hypothesis, float scaleFactor)	Calculates the scaled Tanimoto dissimilarity for floating point values.
static float	scaledTanimoto(int[] v1, int[] v2, float[] hypothesis, float scaleFactor)	Calculates the scaled Tanimoto dissimilarity for floating point values.
static float	tanimoto(float[] v1, float[] v2)	Calculates Tanimoto dissimilarity for floating point values.
static float	tanimoto(int[] v1, int[] v2)	Calculates Tanimoto dissimilarity for floating point values.
static float	tversky(float[] v1, float wv1, float[] v2, float wv2)	Calculates Tversky dissimilarity for floating point values.
static float	tversky(int[] v1, float wv1, int[] v2, float wv2)	Calculates Tversky dissimilarity index for integer values.
static float	weightedAsymmetricEuclidean(float[] v1, float[] v2, float[] weights, float asymmetryFactor)	Calculates the weighted asymmetric Euclidean distance.
static float	weightedAsymmetricEuclidean(int[] v1, int[] v2, float[] weights, float asymmetryFactor)	Calculates the weighted asymmetric Euclidean distance.
static float	weightedAsymmetricNormalizedEuclidean(float[] v1, float[] v2, float[] weights, float asymmetryFactor)	Calculates the weighted asymmetric normalized Euclidean distance.
static float	weightedAsymmetricNormalizedEuclidean(int[] v1, int[] v2, float[] weights, float asymmetryFactor)	Calculates the weighted asymmetric normalized Euclidean distance.
static float	weightedEuclidean(float[] v1, float[] v2, float[] weights)	Calculates the weighted Euclidean distance.
static float	weightedEuclidean(int[] v1, int[] v2, float[] weights)	Calculates the weighted Euclidean distance.
static float	weightedNormalizedEuclidean(float[] v1, float[] v2, float[] weights)	Calculates the weighted normalized Euclidean distance.
static float	weightedNormalizedEuclidean(int[] v1, int[] v2, float[] weights)	Calculates the weighted normalized Euclidean distance.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Details

Metrics

public Metrics()

Method Details

euclidean

public static float euclidean(float[] v1,
 float[] v2)

Calculates the Euclidean distance. The dissimilarity coefficient returned ranges from 0 to MAX_FLOAT, this coefficient is not normalized.

Parameters:

v1 - first vector

v2 - second vector

Returns:

dissimilarity coefficient

normalizedEuclidean

public static float normalizedEuclidean(float[] v1,
 float[] v2)

Calculates the normalized Euclidean distance. The dissimilarity coefficient returned ranges from 0 to 1.

Parameters:

v1 - first vector

v2 - second vector

Returns:

dissimilarity coefficient

asymmetricEuclidean

public static float asymmetricEuclidean(float[] v1,
 float[] v2,
 double asymmetryFactor)

Calculates the asymmetric Euclidean distance. The dissimilarity coefficient returned ranges from 0 to MAX_FLOAT. This metric is also known as directed Euclidean.

Parameters:

v1 - first vector

v2 - second vector

asymmetryFactor - asymmetry ratio

Returns:

dissimilarity coefficient

asymmetricNormalizedEuclidean

public static float asymmetricNormalizedEuclidean(float[] v1,
 float[] v2,
 double asymmetryFactor)

Calculates the asymmetric normalized Euclidean distance. The dissimilarity coefficient returned ranges from 0 to 1.

Parameters:

v1 - first vector

v2 - second vector

asymmetryFactor - asymmetry ratio

Returns:

dissimilarity coefficient

weightedEuclidean

public static float weightedEuclidean(float[] v1,
 float[] v2,
 float[] weights)

Calculates the weighted Euclidean distance.

Parameters:

v1 - first vector

v2 - second vector

weights - weight factors for every coordinate

Returns:

dissimilarity coefficient

weightedNormalizedEuclidean

public static float weightedNormalizedEuclidean(float[] v1,
 float[] v2,
 float[] weights)

Calculates the weighted normalized Euclidean distance.

Parameters:

v1 - first vector

v2 - second vector

weights - weight factors for every coordinate

Returns:

dissimilarity coefficient

weightedAsymmetricEuclidean

public static float weightedAsymmetricEuclidean(float[] v1,
 float[] v2,
 float[] weights,
 float asymmetryFactor)

Calculates the weighted asymmetric Euclidean distance.

Parameters:

v1 - first vector

v2 - second vector

weights - weight factors for every coordinate

asymmetryFactor - asymmetry ratio

Returns:

dissimilarity coefficient

weightedAsymmetricNormalizedEuclidean

public static float weightedAsymmetricNormalizedEuclidean(float[] v1,
 float[] v2,
 float[] weights,
 float asymmetryFactor)

Calculates the weighted asymmetric normalized Euclidean distance.

Parameters:

v1 - first vector

v2 - second vector

weights - weight factors for every coordinate

asymmetryFactor - asymmetry ratio

Returns:

dissimilarity coefficient

tanimoto

public static float tanimoto(float[] v1,
 float[] v2)

Calculates Tanimoto dissimilarity for floating point values.

Parameters:

v1 - first vector

v2 - second vector

Returns:

Tanimoto dissimilarity

tversky

public static float tversky(float[] v1,
 float wv1,
 float[] v2,
 float wv2)

Calculates Tversky dissimilarity for floating point values.

Parameters:

v1 - first vector

wv1 - weight for the first vector

v2 - second vector

wv2 - weight for the second vector

Returns:

Tversky index

asymmetricTanimoto

public static float asymmetricTanimoto(float[] v1,
 float[] v2,
 float asymmetryFactor)

Calculates the asymmetric Tanimoto dissimilarity for floating point values.

Parameters:

v1 - first vector

v2 - second vector

asymmetryFactor - asymmetry ratio

Returns:

Tanimoto dissimilarity

scaledTanimoto

public static float scaledTanimoto(float[] v1,
 float[] v2,
 float[] hypothesis,
 float scaleFactor)

Calculates the scaled Tanimoto dissimilarity for floating point values.

Parameters:

v1 - first vector

v2 - second vector

hypothesis - scaling (consensus) hypothesis

scaleFactor - scale factor

Returns:

Tanimoto dissimilarity

scaledAsymmetricTanimoto

public static float scaledAsymmetricTanimoto(float[] v1,
 float[] v2,
 float[] hypothesis,
 float scaleFactor,
 float asymmetryFactor)

Calculates the scaled asymmetric Tanimoto dissimilarity for floating point values.

Parameters:

v1 - first vector

v2 - second vector

hypothesis - scaling (consensus) hypothesis

scaleFactor - scale factor

asymmetryFactor - asymmetry ratio

Returns:

Tanimoto dissimilarity

euclidean

public static float euclidean(int[] v1,
 int[] v2)

Calculates the Euclidean distance. The dissimilarity coefficient returned ranges from 0 to MAX_FLOAT, this coefficient is not normalized.

Parameters:

v1 - first vector

v2 - second vector

Returns:

dissimilarity coefficient

normalizedEuclidean

public static float normalizedEuclidean(int[] v1,
 int[] v2)

Calculates the normalized Euclidean distance. The dissimilarity coefficient returned ranges from 0 to 1.

Parameters:

v1 - first vector

v2 - second vector

Returns:

dissimilarity coefficient

asymmetricEuclidean

public static float asymmetricEuclidean(int[] v1,
 int[] v2,
 double asymmetryFactor)

Calculates the asymmetric Euclidean distance. The dissimilarity coefficient returned ranges from 0 to MAX_FLOAT. This metric is also known as directed Euclidean.

Parameters:

v1 - first vector

v2 - second vector

asymmetryFactor - asymmetry ratio

Returns:

dissimilarity coefficient

asymmetricNormalizedEuclidean

public static float asymmetricNormalizedEuclidean(int[] v1,
 int[] v2,
 double asymmetryFactor)

Calculates the asymmetric normalized Euclidean distance. The dissimilarity coefficient returned ranges from 0 to 1.

Parameters:

v1 - first vector

v2 - second vector

asymmetryFactor - asymmetry ratio

Returns:

dissimilarity coefficient

weightedEuclidean

public static float weightedEuclidean(int[] v1,
 int[] v2,
 float[] weights)

Calculates the weighted Euclidean distance.

Parameters:

v1 - first vector

v2 - second vector

weights - weight factors for every coordinate

Returns:

dissimilarity coefficient

weightedNormalizedEuclidean

public static float weightedNormalizedEuclidean(int[] v1,
 int[] v2,
 float[] weights)

Calculates the weighted normalized Euclidean distance.

Parameters:

v1 - first vector

v2 - second vector

weights - weight factors for every coordinate

Returns:

dissimilarity coefficient

weightedAsymmetricEuclidean

public static float weightedAsymmetricEuclidean(int[] v1,
 int[] v2,
 float[] weights,
 float asymmetryFactor)

Calculates the weighted asymmetric Euclidean distance.

Parameters:

v1 - first vector

v2 - second vector

weights - weight factors for every coordinate

asymmetryFactor - asymmetry ratio

Returns:

dissimilarity coefficient

weightedAsymmetricNormalizedEuclidean

public static float weightedAsymmetricNormalizedEuclidean(int[] v1,
 int[] v2,
 float[] weights,
 float asymmetryFactor)

Calculates the weighted asymmetric normalized Euclidean distance.

Parameters:

v1 - first vector

v2 - second vector

weights - weight factors for every coordinate

asymmetryFactor - asymmetry ratio

Returns:

dissimilarity coefficient

tanimoto

public static float tanimoto(int[] v1,
 int[] v2)

Calculates Tanimoto dissimilarity for floating point values.

Parameters:

v1 - first vector

v2 - second vector

Returns:

Tanimoto dissimilarity

tversky

public static float tversky(int[] v1,
 float wv1,
 int[] v2,
 float wv2)

Calculates Tversky dissimilarity index for integer values.

Parameters:

v1 - first vector

wv1 - weight for the first vector

v2 - second vector

wv2 - weight for the second vector

Returns:

Tversky dissimilarity index

asymmetricTanimoto

public static float asymmetricTanimoto(int[] v1,
 int[] v2,
 float asymmetryFactor)

Calculates the asymmetric Tanimoto dissimilarity for floating point values.

Parameters:

v1 - first vector

v2 - second vector

asymmetryFactor - asymmetry ratio

Returns:

Tanimoto dissimilarity

scaledTanimoto

public static float scaledTanimoto(int[] v1,
 int[] v2,
 float[] hypothesis,
 float scaleFactor)

Calculates the scaled Tanimoto dissimilarity for floating point values.

Parameters:

v1 - first vector

v2 - second vector

hypothesis - scaling (consensus) hypothesis

scaleFactor - scale factor

Returns:

Tanimoto dissimilarity

scaledAsymmetricTanimoto

public static float scaledAsymmetricTanimoto(int[] v1,
 int[] v2,
 float[] hypothesis,
 float scaleFactor,
 float asymmetryFactor)

Calculates the scaled asymmetric Tanimoto dissimilarity for floating point values.

Parameters:

v1 - first vector

v2 - second vector

hypothesis - scaling (consensus) hypothesis

scaleFactor - scale factor

asymmetryFactor - asymmetry ratio

Returns:

Tanimoto dissimilarity

calcBitCount

public static int calcBitCount(int i)

Counts the number of bits set (1) in the integer parameter.

Parameters:

i - an integer value

Returns:

number of 1 bits in the parameter passed

calcBitCount

public static int calcBitCount(int[] v)

Counts the number of bits set (1) in a bit vector.

Parameters:

v - bit vector, represented by an integer vector, each element stores 32 bits

Returns:

number of 1 bits in the bit vector

binaryEuclidean

public static float binaryEuclidean(int[] v1,
 int[] v2)

Calculates the Euclidean distance of two binary descriptors.

Parameters:

v1 - first vector

v2 - second vector

Returns:

dissimilarity coefficient

binaryNormalizedEuclidean

public static float binaryNormalizedEuclidean(int[] v1,
 int[] v2)

Calculates the normalized Euclidean distance of two binary descriptors.

Parameters:

v1 - first vector

v2 - second vector

Returns:

dissimilarity coefficient

binaryNormalizedEuclidean

public static float binaryNormalizedEuclidean(int[] v1,
 int oneCount1,
 int[] v2,
 int oneCount2)

Calculates the normalized Euclidean distance of two binary descriptors. This method should be called in place of binaryNormalizedEuclidean( final int[] v1, final int[] v2 ) when the number of bits set (1) is known in both input vectors, as this method does not recalculate these values which results in faster operation.

Parameters:

v1 - first vector

oneCount1 - number of 1 bits in v1

v2 - second vector

oneCount2 - number of 1 bits in v1

Returns:

dissimilarity coefficient

binaryWeightedEuclidean

public static float binaryWeightedEuclidean(int[] v1,
 int[] v2,
 float[] weights)

Calculates the weighted Euclidean distance of binary descriptors.

Parameters:

v1 - first vector

v2 - second vector

weights - weight factors for every coordinate

Returns:

dissimilarity coefficient

binaryWeightedNormalizedEuclidean

public static float binaryWeightedNormalizedEuclidean(int[] v1,
 int[] v2,
 float[] weights)

Calculates the weighted normalized Euclidean distance of binary descriptors.

Parameters:

v1 - first vector

v2 - second vector

weights - weight factors for every coordinate

Returns:

dissimilarity coefficient

binaryWeightedNormalizedEuclidean

public static float binaryWeightedNormalizedEuclidean(int[] v1,
 int oneCount1,
 int[] v2,
 int oneCount2,
 int[] weights)

Calculates the weighted normalized Euclidean distance of binary descriptors.

Parameters:

v1 - first vector

oneCount1 - number of 1 bits in v1

v2 - second vector

oneCount2 - number of 1 bits in v2

weights - weight factors for every coordinate

Returns:

dissimilarity coefficient

binaryAsymmetricEuclidean

public static float binaryAsymmetricEuclidean(int[] v1,
 int[] v2,
 float asymmetryFactor)

Calculates the asymmetric Euclidean distance of binary descriptors.

Parameters:

v1 - first vector

v2 - second vector

asymmetryFactor - asymmetry ratio

Returns:

dissimilarity coefficient

binaryAsymmetricNormalizedEuclidean

public static float binaryAsymmetricNormalizedEuclidean(int[] v1,
 int[] v2,
 float asymmetryFactor)

Calculates the asymmetric normalized Euclidean distance of binary descriptors.

Parameters:

v1 - first vector

v2 - second vector

asymmetryFactor - asymmetry ratio

Returns:

dissimilarity coefficient

binaryAsymmetricNormalizedEuclidean

public static float binaryAsymmetricNormalizedEuclidean(int[] v1,
 int oneCount1,
 int[] v2,
 int oneCount2,
 float asymmetryFactor)

Calculates the asymmetric normalized Euclidean distance of binary descriptors. This method should be called in place of binaryAsymmetricNormalizedEuclidean( final int[] v1, final int[] v2 ) when the number of bits set (1) is known in both input vectors, as this method does not recalculate these values which results in faster operation.

Parameters:

v1 - first vector

oneCount1 - number of 1 bits in v1

v2 - second vector

oneCount2 - number of 1 bits in v2

asymmetryFactor - asymmetry ratio

Returns:

dissimilarity coefficient

binaryWeightedAsymmetricEuclidean

public static float binaryWeightedAsymmetricEuclidean(int[] v1,
 int[] v2,
 float[] weights,
 float asymmetryFactor)

Calculates the weighted asymmetric Euclidean distance of binary descriptors.

Parameters:

v1 - first vector

v2 - second vector

weights - weight factors for every coordinate

asymmetryFactor - asymmetry ratio

Returns:

dissimilarity coefficient

binaryWeightedAsymmetricNormalizedEuclidean

public static float binaryWeightedAsymmetricNormalizedEuclidean(int[] v1,
 int[] v2,
 float[] weights,
 float asymmetryFactor)

Calculates the weighted asymmetric normalized Euclidean distance of binary descriptors.

Parameters:

v1 - first vector

v2 - second vector

weights - weight factors for every coordinate

asymmetryFactor - asymmetry ratio

Returns:

dissimilarity coefficient

binaryWeightedAsymmetricNormalizedEuclidean

public static float binaryWeightedAsymmetricNormalizedEuclidean(int[] v1,
 int oneCount1,
 int[] v2,
 int oneCount2,
 float[] weights,
 float asymmetryFactor)

Calculates the weighted asymmetric normalized Euclidean distance of binary descriptors. This method should be called in place of binaryWeightedAsymmetricNormalizedEuclidean( final int[] v1, final int[] v2, final float[] weights, float asymmetryFactor ) when the number of bits set (1) is known in both input vectors, as this method does not recalculate these values which results in faster operation.

Parameters:

v1 - first vector

oneCount1 - number of 1 bits in v1

v2 - second vector

oneCount2 - number of 1 bits in v2

weights - weight factors for every coordinate

asymmetryFactor - asymmetry ratio

Returns:

dissimilarity coefficient

binaryTanimoto

public static float binaryTanimoto(int[] v1,
 int[] v2)

Calculates Tanimoto dissimilarity for binary descriptors.

Parameters:

v1 - first vector

v2 - second vector

Returns:

Tanimoto dissimilarity

binaryTversky

public static float binaryTversky(int[] v1,
 float wv1,
 int[] v2,
 float wv2)

Calculates Tversky dissimilarity for binary descriptors.

Parameters:

v1 - first vector

wv1 - weight for the first vector

v2 - second vector

wv2 - weight for the second vector

Returns:

Tversky dissimilarity

binaryAsymmetricTanimoto

public static float binaryAsymmetricTanimoto(int[] v1,
 int[] v2,
 float asymmetryFactor)

Calculates asymmetric Tanimoto dissimilarity for binary descriptors.

Parameters:

v1 - first vector

v2 - second vector

asymmetryFactor - asymmetry ratio

Returns:

Tanimoto dissimilarity

binaryTanimoto

public static float binaryTanimoto(int[] v1,
 int oneCount1,
 int[] v2,
 int oneCount2)

Calculates Tanimoto dissimilarity for binary descriptors. This method should be called in place of binaryTanimoto( final int[] v1, final int[] v2 ) when the number of bits set (1) is known in both input vectors, as this method does not recalculate these values which results in faster operation.

Parameters:

v1 - first vector

oneCount1 - number of 1 bits in v1

v2 - second vector

oneCount2 - number of 1 bits in v2

Returns:

Tanimoto dissimilarity

binaryAsymmetricTanimoto

public static float binaryAsymmetricTanimoto(int[] v1,
 int oneCount1,
 int[] v2,
 int oneCount2,
 float asymmetryFactor)

Calculates asymmetricTanimoto dissimilarity for binary descriptors. This method should be called in place of binaryTanimoto( final int[] v1, final int[] v2, float asymmetryFactor ) when the number of bits set (1) is known in both input vectors, as this method does not recalculate these values which results in faster operation.

Parameters:

v1 - first vector

oneCount1 - number of 1 bits in v1

v2 - second vector

oneCount2 - number of 1 bits in v2

asymmetryFactor - asymmetry ratio

Returns:

Tanimoto dissimilarity