OpenSwathPeptidoformInference Class Reference

OpenSwath inference of peptidoforms (IPF). More...

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

Classes
struct	BayesianModelRow
	Compact Bayesian model row used by the public helper methods. More...
struct	PosteriorRow
	Posterior probability row returned by applyBM. More...

Public Member Functions
std::vector< IPFPrecursorProbabilityRow >	precursorInference (const std::vector< IPFPrecursorRow > &precursor_rows, const PeptidoformInferenceConfig &config) const
	Perform precursor-layer inference and return precursor peakgroup PEPs.
std::vector< IPFResultRow >	infer (const std::vector< IPFPrecursorRow > &precursor_rows, const std::vector< IPFTransitionRow > &transition_rows, const std::vector< IPFAlignmentRow > &alignment_rows, const PeptidoformInferenceConfig &config) const
	Perform complete peptidoform inference.

Static Public Member Functions
static std::vector< double >	computeModelFDR (const std::vector< double > &pep_values)
	Compute model FDR/q-values from posterior error probabilities.
static std::vector< BayesianModelRow >	preparePrecursorBM (const std::vector< IPFPrecursorRow > &rows)
	Build precursor-layer Bayesian model rows.
static std::vector< PosteriorRow >	applyBM (const std::vector< BayesianModelRow > &rows)
	Apply a compact Bayesian model and return posterior probabilities.

Detailed Description

OpenSwath inference of peptidoforms (IPF).

This class is file-format agnostic and works only on compact typed rows.

See: Rosenberger, G., Liu, Y., Röst, H. et al. Inference and quantification of peptidoforms in large sample cohorts by SWATH-MS. Nat Biotechnol 35, 781–788 (2017). https://doi.org/10.1038/nbt.3908

Member Function Documentation

applyBM()

static std::vector< PosteriorRow > applyBM ( const std::vector< BayesianModelRow > &  rows )

static

Apply a compact Bayesian model and return posterior probabilities.

Rows are grouped by feature and hypothesis. Within each group, evidence values are multiplied and the smallest prior in the group is retained, matching the PyProphet implementation being ported.

Parameters

rows	Bayesian-model rows

Returns

Posterior probabilities per feature/hypothesis pair

computeModelFDR()

static std::vector< double > computeModelFDR ( const std::vector< double > &  pep_values )

static

Compute model FDR/q-values from posterior error probabilities.

This reuses OpenMS multiple-testing utilities and follows the PyProphet IPF convention of computing model FDR from sorted PEP values with maximum-rank tie handling.

Parameters

pep_values

Posterior error probabilities for one model-FDR family

Returns

Model FDR / q-values in the original input order

infer()

std::vector< IPFResultRow > infer	(	const std::vector< IPFPrecursorRow > &	precursor_rows,
		const std::vector< IPFTransitionRow > &	transition_rows,
		const std::vector< IPFAlignmentRow > &	alignment_rows,
		const PeptidoformInferenceConfig &	config
	)		const

Perform complete peptidoform inference.

Parameters

precursor_rows	Peakgroup/precursor evidence
transition_rows	Transition-level evidence
alignment_rows	Optional alignment-group memberships for across-run propagation
config	IPF configuration

Returns

Final IPF result rows without the H0 hypothesis

precursorInference()

std::vector< IPFPrecursorProbabilityRow > precursorInference	(	const std::vector< IPFPrecursorRow > &	precursor_rows,
		const PeptidoformInferenceConfig &	config
	)		const

Perform precursor-layer inference and return precursor peakgroup PEPs.

This step combines peakgroup evidence with optional MS1/MS2 precursor evidence. The returned precursor peakgroup PEP is then used as the prior for the transition/peptidoform layer.

Parameters

precursor_rows	Peakgroup/precursor evidence rows
config	IPF configuration

Returns

One precursor peakgroup PEP per retained feature

preparePrecursorBM()

static std::vector< BayesianModelRow > preparePrecursorBM ( const std::vector< IPFPrecursorRow > &  rows )

static

Build precursor-layer Bayesian model rows.

For each feature this emits rows for the true (1) and false (0) precursor hypotheses using the configured MS1/MS2 precursor evidence.

Parameters

rows	Peakgroup and optional precursor-evidence rows

Returns

Bayesian-model rows for precursor inference