Generic namespace for all classes and functions of DIAlign. More...

Namespaces
	AffineAlignment
	Includes functions for the affine-alignment through a score-matrix using gap-opening and gap-closing penalties.

	Alignment
	Contains some of the core alignment functions for non-affine alignment.

	ConstrainMatrix
	Namespace for constraing similarity score matrix.

	SimilarityMatrix
	Similarity matrix namespace.

	Traceback
	Generic namespace for Traceback that includes source matrix name and the arrow direction.

	Utils
	Utility namespace.

Classes
struct	AffineAlignObj
	An affine alignment object. More...

struct	AlignedIndices
	Includes aligned indices of signal A and signal B with the cumulative score. More...

struct	AlignObj
	An alignment object (for non-affine alignment) More...

struct	SimMatrix
	Similarity matrix. More...

struct	SimMatrix_bool
	Path matrix. More...

Functions
void	alignChromatogramsCpp (AffineAlignObj &obj, const std::vector< std::vector< double > > &r1, const std::vector< std::vector< double > > &r2, std::string alignType, const std::vector< double > &tA, const std::vector< double > &tB, const std::string &normalization, const std::string &simType, double B1p=0.0, double B2p=0.0, int noBeef=0, double goFactor=0.125, double geFactor=40, double cosAngleThresh=0.3, bool OverlapAlignment=true, double dotProdThresh=0.96, double gapQuantile=0.5, bool hardConstrain=false, double samples4gradient=100.0)
	Align two pairs of chromatograms derived from two LC-MS/MS experiments A and B. More...

SimMatrix	getSimilarityMatrix (const std::vector< std::vector< double >> &d1, const std::vector< std::vector< double >> &d2, const std::string &Normalization, const std::string &SimType, double cosAngleThresh, double dotProdThresh)
	Calculates similarity matrix of two fragment-ion chromatogram groups or extracted-ion chromatograms (XICs) derived from two LC-MS/MS experiments A and B. More...

void	doAffineAlignment (AffineAlignObj &obj, const SimMatrix &s, double goPenalty, double gePenalty, bool OverlapAlignment)
	Performs affine alignment on a given alignment object. More...

void	getAffineAlignedIndices (AffineAlignObj &obj, int bandwidth=0)
	Compute path using AffineAlignObj. More...

double	getGapPenalty (const SimMatrix &s, double gapQuantile, std::string SimType)
	returns a gap penalty from the distribution of similarity scores. More...

double	areaBwBoundaries (std::vector< std::vector< double > > vov, int leftIdx, int rightIdx)
	returns the summation of signals between leftIdx and rightIdx from vov.

Variables
const int	NA = 0

Detailed Description

Generic namespace for all classes and functions of DIAlign.

This contains a brief description of the available functions in DIAlign. The DIAlign algorithm can be called through the interface provided by alignChromatogramsCpp(), which allows alignment of a set of chromatograms.

std::vector< std::vector<double > > data1, data2; // chrom coordinates
std::vector<double > rt_data1, data2; // RT coordinates
std::vector<double > rt_mapping; // global RT alignment
int noBeef = 5; // nr of matrix cells w/o penalty
AffineAlignObj alignment_obj(rt_data1.size(), rt_data1.size(), false); // Initialize AffineAlignObj
alignChromatogramsCpp(alignment_obj,
  data1, data2,
  "hybrid",
  rt_data1, rt_data2,
  "mean", "dotProductMasked",
  rt_mapping.front(), rt_mapping.back(),
  noBeef);

The AffineAlignObj will have the fields indexA_aligned, indexB_aligned and score populated, providing the aligned indices as output.

Individual functions that perform the alignment can be called through getSimilarityMatrix(), doAffineAlignment(), getAffineAlignedIndices().

Function Documentation

void DIAlign::alignChromatogramsCpp	(	AffineAlignObj &	obj,
		const std::vector< std::vector< double > > &	r1,
		const std::vector< std::vector< double > > &	r2,
		std::string	alignType,
		const std::vector< double > &	tA,
		const std::vector< double > &	tB,
		const std::string &	normalization,
		const std::string &	simType,
		double	B1p = `0.0`,
		double	B2p = `0.0`,
		int	noBeef = `0`,
		double	goFactor = `0.125`,
		double	geFactor = `40`,
		double	cosAngleThresh = `0.3`,
		bool	OverlapAlignment = `true`,
		double	dotProdThresh = `0.96`,
		double	gapQuantile = `0.5`,
		bool	hardConstrain = `false`,
		double	samples4gradient = `100.0`
	)

Align two pairs of chromatograms derived from two LC-MS/MS experiments A and B.

Parameters

obj	A object of type AffineAlignObj which needs to have at least capacity of the chromatogram length
r1	The first set of chromatograms, where each chromatogram is a vector of intensities. These chromatograms are XICs derived from a single analyte in run A
r2	The second set of chromatograms, where each chromatogram is a vector of intensities. These chromatograms are XICs derived from a single analyte in run B
alignType	Available alignment methods, must be one of "global", "local" and "hybrid".
tA	This vector has equally spaced timepoints of the XICs from run A.
tB	This vector has equally spaced timepoints of the XICs from run B.
normalization	Normalization method, must be one of "L2", "mean" or "none".
simType	Similarity computation method, defines how similarity between two chromatographic points is computed. Must be one of (dotProductMasked, dotProduct, cosineAngle, cosine2Angle, euclideanDist, covariance, correlation).
B1p	Timepoint mapped by global fit for tA[0].
B2p	Timepoint mapped by global fit for tA[tA.size()-1].
noBeef	It defines the distance from the global fit, upto which no penalization is performed. The window length will be chosen as = 2*noBeef.
goFactor	Penalty for introducing first gap in alignment. This value is multiplied by base gap-penalty.
geFactor	Penalty for introducing subsequent gaps in alignment. This value is multiplied by base gap-penalty.
cosAngleThresh	In simType == dotProductMasked mode, angular similarity should be higher than cosAngleThresh otherwise similarity is forced to zero.
OverlapAlignment	If alignment should be performed with free end-gaps. False: Global alignment, True: overlap alignment.
dotProdThresh	In simType = dotProductMasked mode, values in similarity matrix higher than dotProdThresh quantile are checked for angular similarity.
gapQuantile	Must be between 0 and 1. This is used to calculate base gap-penalty from similarity distribution.
hardConstrain	Whether a hard constraint should be used for the similarity matrix. If false; indices farther from noBeef distance are filled with distance from linear fit line.
samples4gradient	This parameter modulates penalization of masked indices.

Note: r1 and r2 need to have the same length and the chromatograms need to be in the same order (e.g. r1[0] needs to be the same XIC trace as r2[0]); If no global fit is available, it is recommended to set B1p to tB.front() and B2p to tB.back()

void DIAlign::doAffineAlignment	(	AffineAlignObj &	obj,
		const SimMatrix &	s,
		double	goPenalty,
		double	gePenalty,
		bool	OverlapAlignment
	)

Performs affine alignment on a given alignment object.

Note: Does not perform back-tracking, please call getAffineAlignedIndices() afterwards.

Parameters

obj	A object of type AffineAlignObj which needs to have at least capacity of the chromatogram length
s	A previously computed similarity matrix
goPenalty	Penalty for introducing first gap in alignment.
gePenalty	Penalty for introducing subsequent gaps in alignment.
OverlapAlignment	If alignment should be performed with free end-gaps. False: Global alignment, True: overlap alignment.

void DIAlign::getAffineAlignedIndices	(	AffineAlignObj &	obj,
		int	bandwidth = `0`
	)

Compute path using AffineAlignObj.

Parameters

obj	A object of type AffineAlignObj which needs to have matrices A, B and M filled.
bandwith	Compute path for multiple matrix cells (bandwith is number of cells to be computed)

Note: this assumes that matrices A, B and M have been computed. Use after calling doAffineAlignment().; this will compute the following members: indexA_aligned, indexB_aligned, score.

double DIAlign::getGapPenalty	(	const SimMatrix &	s,
		double	gapQuantile,
		std::string	SimType
	)

returns a gap penalty from the distribution of similarity scores.

For simType = "cosineAngle" and "cosine2Angle", gap penalty = 0.95 Other simType values, gap Penalty is the gapQuantile quantile from the score disctribution of s. gapQuantile must be between 0 and 1.

SimMatrix DIAlign::getSimilarityMatrix	(	const std::vector< std::vector< double >> &	d1,
		const std::vector< std::vector< double >> &	d2,
		const std::string &	Normalization,
		const std::string &	SimType,
		double	cosAngleThresh,
		double	dotProdThresh
	)

Calculates similarity matrix of two fragment-ion chromatogram groups or extracted-ion chromatograms (XICs) derived from two LC-MS/MS experiments A and B.

Parameters

d1	The first set of chromatograms, where each chromatogram is a vector of intensities. These chromatograms are XICs derived from a single analyte in run A
d2	The second set of chromatograms, where each chromatogram is a vector of intensities. These chromatograms are XICs derived from a single analyte in run B
Normalization	Normalization method, must be one of "L2", "mean" or "none".
SimType	Similarity computation method, defines how similarity between two chromatographic points is computed. Must be one of (dotProductMasked, dotProduct, cosineAngle, cosine2Angle, euclideanDist, covariance, correlation).
cosAngleThresh	In simType == dotProductMasked mode, angular similarity should be higher than cosAngleThresh otherwise similarity is forced to zero.
dotProdThresh	In simType = dotProductMasked mode, values in similarity matrix higher than dotProdThresh quantile are checked for angular similarity.

Returns: Returns a similarity matrix

Namespaces

Classes

Functions

Variables

Detailed Description

Function Documentation