Confidence scoring for SRM/MRM/PRM features against a targeted assay library. More...

#include <OpenMS/ANALYSIS/OPENSWATH/ConfidenceScoring.h>

Inheritance diagram for ConfidenceScoring:

Collaboration diagram for ConfidenceScoring:

Classes
struct	GLM_
	Binomial GLM used to map (squared-normalised-RT-diff, intensity-distance) -> [0, 1] confidence. More...

struct	RTNorm_
	Map RT values into the [0, 100] interval using min/max RT of the assay library. More...

Public Member Functions
	ConfidenceScoring (bool test_mode_=false)
	Construct an empty scorer.

	~ConfidenceScoring () override
	Destructor.

void	initialize (const TargetedExperiment &library, const Size n_decoys, const Size n_transitions, const TransformationDescription &rt_trafo)
	Install the configuration needed before scoreMap() can run.

void	initializeGlm (double intercept, double rt_coef, double int_coef)
	Install the GLM coefficients fitted externally on a training set.

void	scoreMap (FeatureMap &features)
	Score every feature in `features` in place, writing per-feature scores and an updated overall-quality value.

Public Member Functions inherited from ProgressLogger
	ProgressLogger ()
	Constructor.

virtual	~ProgressLogger ()
	Destructor.

	ProgressLogger (const ProgressLogger &other)
	Copy constructor.

ProgressLogger &	operator= (const ProgressLogger &other)
	Assignment Operator.

void	setLogType (LogType type) const
	Sets the progress log that should be used. The default type is NONE!

LogType	getLogType () const
	Returns the type of progress log being used.

void	setLogger (ProgressLoggerImpl *logger)
	Sets the logger to be used for progress logging.

void	startProgress (SignedSize begin, SignedSize end, const std::string &label) const
	Initializes the progress display.

void	setProgress (SignedSize value) const
	Sets the current progress.

void	endProgress (UInt64 bytes_processed=0) const

void	nextProgress () const
	increment progress by 1 (according to range begin-end)

Protected Member Functions
void	chooseDecoys_ ()
	Permute `decoy_index_` in place to pick a fresh random decoy sample for the next feature.

double	manhattanDist_ (DoubleList x, DoubleList y)
	Manhattan (L1) distance between two equal-length vectors.

double	getAssayRT_ (const TargetedExperiment::Peptide &assay)
	Read the (single) retention time from an assay's TargetedExperiment::Peptide; the assay is required to have an RT set (checked in debug builds)

double	scoreAssay_ (const TargetedExperiment::Peptide &assay, double feature_rt, DoubleList &feature_intensities, const std::set< std::string > &transition_ids=std::set< std::string >())
	Score one feature against one candidate assay.

void	scoreFeature_ (Feature &feature)
	Score one feature against its matching assay plus a random decoy sample; writes the per-assay scores into `feature's` MetaValues.

Protected Attributes
struct OpenMS::ConfidenceScoring::GLM_	glm_

struct OpenMS::ConfidenceScoring::RTNorm_	rt_norm_

TargetedExperiment	library_
	Targeted-assay library: one peptide per assay, each with its transitions.

IntList	decoy_index_
	Indexes into `library_.getPeptides()` used as decoys for the current feature.

Size	n_decoys_
	Number of decoy assays to sample per feature (0 = use all unrelated assays as decoys)

std::map< std::string, IntList >	transition_map_
	Lookup assay-id -> indexes into `library_.getTransitions()`

Size	n_transitions_
	Number of top-intensity transitions to keep when computing the intensity-distance term (0 = keep all)

TransformationDescription	rt_trafo_
	Optional RT transformation applied to measured feature RTs before comparison with library RTs.

Math::RandomShuffler	shuffler_
	Random shuffler used to draw decoy samples (seed depends on test mode — see ctor)

Protected Attributes inherited from ProgressLogger
LogType	type_

time_t	last_invoke_

ProgressLoggerImpl *	current_logger_

Additional Inherited Members
Public Types inherited from ProgressLogger
enum	LogType { CMD , GUI , NONE }
	Possible log types. More...

Static Protected Attributes inherited from ProgressLogger
static int	recursion_depth_

Detailed Description

Confidence scoring for SRM/MRM/PRM features against a targeted assay library.

For each detected feature the algorithm scores the feature against

the matching true assay (referenced by the feature's PeptideRef MetaValue), and
a random sample of decoy assays (randomly chosen non-matching assays from the same library) used to estimate the score distribution under the null hypothesis.

The per-assay score is the output of a binomial GLM (see GLM_) on two features derived from the feature/assay pair:

the difference between the feature's RT and the assay's library RT, after normalisation to the 0–100 range (see RTNorm_); the GLM itself applies the square (its rt_coef multiplies diff_rt*diff_rt).
the Manhattan distance between the feature's transition-intensity vector and the assay's expected transition-intensity vector (intensities matched by Q3 m/z).

Initialisation flow:

initialize() — install the library, decoy count, transition count, and RT transform.
initializeGlm() — install the pre-fitted GLM coefficients (intercept / RT² / intensity).
scoreMap() — iterate all features in the input FeatureMap and write the per-feature scoring result back (see scoreMap() for the exact MetaValue set + overall-quality update).

Feature → assay matching is by MetaValue: each feature must carry a "PeptideRef" string (the assay id), and each subordinate (one per transition) must carry a "native_id" string matching a transition id from the library.

Constructor & Destructor Documentation

◆ ConfidenceScoring()

ConfidenceScoring ( bool test_mode_ = false )

explicit

Construct an empty scorer.

Parameters

[in] test_mode_ If true, seed the internal random shuffler with time(NULL) so that decoy selection is deterministic only within a single process run. If false (default), seed with 0 for fully reproducible decoy draws across runs and platforms.

◆ ~ConfidenceScoring()

~ConfidenceScoring ( )

inlineoverride

Destructor.

Member Function Documentation

◆ chooseDecoys_()

void chooseDecoys_ ( )

protected

Permute decoy_index_ in place to pick a fresh random decoy sample for the next feature.

◆ getAssayRT_()

double getAssayRT_ ( const TargetedExperiment::Peptide & assay )

protected

Read the (single) retention time from an assay's TargetedExperiment::Peptide; the assay is required to have an RT set (checked in debug builds)

◆ initialize()

void initialize	(	const TargetedExperiment &	library,
		const Size	n_decoys,
		const Size	n_transitions,
		const TransformationDescription &	rt_trafo
	)

inline

Install the configuration needed before scoreMap() can run.

Parameters

[in]	library	Targeted-assay library; must have at least 2 peptides for scoring to be defined.
[in]	n_decoys	Number of decoy assays to sample per feature (clamped to all-available if larger, with warning).
[in]	n_transitions	Number of top-intensity transitions to keep when computing the intensity-distance term; 0 keeps all.
[in]	rt_trafo	RT transformation applied to feature RTs before comparison against library RTs.

◆ initializeGlm()

void initializeGlm	(	double	intercept,
		double	rt_coef,
		double	int_coef
	)

inline

Install the GLM coefficients fitted externally on a training set.

Coefficients correspond to the GLM_ formula \(\mathrm{sigmoid}(\mathrm{intercept} + \mathrm{rt\_coef} \cdot \Delta\mathrm{RT}^2 + \mathrm{int\_coef} \cdot d_\mathrm{int})\).

Parameters

[in]	intercept	GLM intercept term.
[in]	rt_coef	Coefficient on \(\Delta\mathrm{RT}^2\) (using normalised RT in [0, 100]).
[in]	int_coef	Coefficient on the Manhattan intensity distance.

◆ manhattanDist_()

double manhattanDist_	(	DoubleList	x,
		DoubleList	y
	)

protected

Manhattan (L1) distance between two equal-length vectors.

◆ scoreAssay_()

double scoreAssay_	(	const TargetedExperiment::Peptide &	assay,
		double	feature_rt,
		DoubleList &	feature_intensities,
		const std::set< std::string > &	transition_ids = `std::set< std::string >()`
	)

protected

Score one feature against one candidate assay.

Reduces the assay's transitions to those listed in transition_ids (or all if the set is empty), matches them up with the feature's transition intensities by Q3 m/z, computes the Manhattan distance on the resulting paired intensity vectors, evaluates the GLM on (RT² diff, intensity distance) and returns the GLM output in [0, 1].

Parameters

[in]	assay	Candidate assay (true or decoy).
[in]	feature_rt	Feature retention time after `rt_trafo_` and `rt_norm_`.
[in,out]	feature_intensities	Feature's transition intensities; reordered to match the assay's transition list.
[in]	transition_ids	Optional subset of transition ids to consider; empty means "use all transitions of the assay".

Returns: GLM-derived score in [0, 1].

◆ scoreFeature_()

void scoreFeature_ ( Feature & feature )

protected

Score one feature against its matching assay plus a random decoy sample; writes the per-assay scores into feature's MetaValues.

◆ scoreMap()

void scoreMap ( FeatureMap & features )

inline

Score every feature in features in place, writing per-feature scores and an updated overall-quality value.

Prerequisite: both initialize and initializeGlm must have been called first.

The input contract is:

features each carry the assay id in MetaValue "PeptideRef".
Each feature has one Feature::getSubordinates entry per assay transition, each carrying the transition id in MetaValue "native_id".
The targeted-assay library initialize() was called with must contain at least 2 peptides; the function throws Exception::IllegalArgument otherwise.

Per-feature side effects:

MetaValue "GLM_score" is set to the GLM score against the true assay.
MetaValue "local_FDR" is set to the rank-based local FDR estimated against the per-feature decoy scores.
The feature's overall quality is set to 1.0 - local_FDR.

Other side effects:

The library's RT range is computed here and cached in rt_norm_.
If n_decoys_ exceeds the number of unrelated assays in the library, it is clamped to 0 (= use all available assays as decoys) and a warning is logged.

Parameters

[in,out] features Feature map to score; per-feature MetaValues and overall quality are written in place.

Exceptions

Exception::IllegalArgument If library_ has fewer than 2 peptides.

References ExposedVector< VectorElement >::begin(), ExposedVector< VectorElement >::end(), TargetedExperiment::getPeptides(), TargetedExperiment::getTransitions(), OPENMS_LOG_DEBUG, OPENMS_LOG_WARN, and ExposedVector< VectorElement >::size().

Member Data Documentation

◆ decoy_index_

IntList decoy_index_

protected

Indexes into library_.getPeptides() used as decoys for the current feature.

◆ glm_

struct OpenMS::ConfidenceScoring::GLM_ glm_

protected

◆ library_

TargetedExperiment library_

protected

Targeted-assay library: one peptide per assay, each with its transitions.

◆ n_decoys_

Size n_decoys_

protected

Number of decoy assays to sample per feature (0 = use all unrelated assays as decoys)

◆ n_transitions_

Size n_transitions_

protected

Number of top-intensity transitions to keep when computing the intensity-distance term (0 = keep all)

◆ rt_norm_

struct OpenMS::ConfidenceScoring::RTNorm_ rt_norm_

protected

◆ rt_trafo_

TransformationDescription rt_trafo_

protected

Optional RT transformation applied to measured feature RTs before comparison with library RTs.

◆ shuffler_

Math::RandomShuffler shuffler_

protected

Random shuffler used to draw decoy samples (seed depends on test mode — see ctor)

◆ transition_map_

std::map<std::string, IntList> transition_map_

protected

Lookup assay-id -> indexes into library_.getTransitions()

Classes

Public Member Functions

Protected Member Functions

Protected Attributes

Additional Inherited Members

Detailed Description

Constructor & Destructor Documentation

◆ ConfidenceScoring()

◆ ~ConfidenceScoring()

Member Function Documentation

◆ chooseDecoys_()

◆ getAssayRT_()

◆ initialize()

◆ initializeGlm()

◆ manhattanDist_()

◆ scoreAssay_()

◆ scoreFeature_()

◆ scoreMap()

Member Data Documentation

◆ decoy_index_

◆ glm_

◆ library_

◆ n_decoys_

◆ n_transitions_

◆ rt_norm_

◆ rt_trafo_

◆ shuffler_

◆ transition_map_