LocalLinearMap Class Reference

Trained Local Linear Map (LLM) model for peak intensity prediction. More...

#include <OpenMS/ANALYSIS/PIP/LocalLinearMap.h>

Classes
struct	LLMParam
	Define parameters needed by the Local Linear Map (LLM) model. More...

Public Member Functions
	LocalLinearMap ()
	default constructor More...
virtual	~LocalLinearMap ()
	destructor More...
const LLMParam &	getLLMParam () const
	return parameters of the LocalLinearMap model More...
const Matrix< double > &	getCodebooks () const
	return position of the codebook vectors (18-dim) More...
const Matrix< double > &	getMatrixA () const
	return linear mappings of the codebooks More...
const std::vector< double > &	getVectorWout () const
	return linear bias More...
const Matrix< UInt > &	getCord () const
	return coordinates of codebook vectors on the 2-d grid More...
void	normalizeVector (std::vector< double > &aaIndexVariables)
	calculate and return normalized amino acid index variables from string representation of peptide More...
std::vector< double >	neigh (const Matrix< UInt > &cord, Size win, double radius)
	calculate neighborhood function based on distance of prototypes to winner prototype on two-dimensional grid structure and neighborhood width. More...

Private Member Functions
Matrix< UInt >	genCord_ (Size xdim, Size ydim)
	needed to store prototype coordinates More...
double	dist_ (const Matrix< UInt > &u, const Matrix< UInt > &v, Size a, Size b)
	calculate distance between two prototypes More...
	LocalLinearMap (LocalLinearMap &rhs)
	Copy constructor not implemented => private. More...
LocalLinearMap &	operator= (const LocalLinearMap &llm)
	Assignment operator not implemented => private. More...

Private Attributes
LLMParam	param_
	parameters of the model More...
Matrix< double >	code_
	codebook vectors More...
Matrix< double >	A_
	linear mappings More...
std::vector< double >	wout_
	linear bias More...
Matrix< UInt >	cord_
	coordinates of codebooks on grid More...

Detailed Description

Trained Local Linear Map (LLM) model for peak intensity prediction.

This class offers a model for predictions of peptide peak heights (referred to as intensities) by a Local Linear Map (LLM) model and is the basis of PeakIntensityPredictor.

A general introduction to the Peak Intensity Predictor (PIP) can be found in the PIP Tutorial.

The model trained needs two files for storing the position of the codebook vectors and the linear mappings (codebooks.data, linearMapping.data) This is the default model used by PeakIntensityPredictor.

---

Class Documentation

OpenMS::LocalLinearMap::LLMParam

struct OpenMS::LocalLinearMap::LLMParam

Define parameters needed by the Local Linear Map (LLM) model.

Parameters xdim and ydim define the size of the two dimensional grid structure. Parameter radius gives the width of the Gaussian neighborhood function.

Class Members
double	radius	width of Gaussian neighborhood function
UInt	xdim	size of first coordinate
UInt	ydim	size of second coordinate

Constructor & Destructor Documentation

LocalLinearMap() [1/2]

LocalLinearMap

(

)

default constructor

~LocalLinearMap()

virtual ~LocalLinearMap ( )

virtual

destructor

LocalLinearMap() [2/2]

LocalLinearMap ( LocalLinearMap &  rhs )

private

Copy constructor not implemented => private.

Member Function Documentation

dist_()

double dist_ ( const Matrix< UInt > &  u, const Matrix< UInt > &  v, Size  a, Size  b  )

private

calculate distance between two prototypes

genCord_()

Matrix<UInt> genCord_ ( Size  xdim, Size  ydim  )

private

needed to store prototype coordinates

getCodebooks()

const Matrix<double>& getCodebooks

(

)

const

return position of the codebook vectors (18-dim)

getCord()

const Matrix<UInt>& getCord

(

)

const

return coordinates of codebook vectors on the 2-d grid

getLLMParam()

const LLMParam& getLLMParam

(

)

const

return parameters of the LocalLinearMap model

getMatrixA()

const Matrix<double>& getMatrixA

(

)

const

return linear mappings of the codebooks

getVectorWout()

const std::vector<double>& getVectorWout

(

)

const

return linear bias

neigh()

std::vector<double> neigh	(	const Matrix< UInt > &	cord,
		Size	win,
		double	radius
	)

calculate neighborhood function based on distance of prototypes to winner prototype on two-dimensional grid structure and neighborhood width.

normalizeVector()

void normalizeVector

(

std::vector< double > &

aaIndexVariables

)

calculate and return normalized amino acid index variables from string representation of peptide

operator=()

LocalLinearMap& operator= ( const LocalLinearMap &  llm )

private

Assignment operator not implemented => private.

Member Data Documentation

A_

Matrix<double> A_

private

linear mappings

code_

Matrix<double> code_

private

codebook vectors

cord_

Matrix<UInt> cord_

private

coordinates of codebooks on grid

param_

LLMParam param_

private

parameters of the model

wout_

std::vector<double> wout_

private

linear bias