|
OpenMS
2.5.0
|
Go to the documentation of this file.
35 #ifndef OPENMS_TRANSFORMATIONS_FEATUREFINDER_FEATUREFINDERIDENTIFICATIONALGORITHM_H
36 #define OPENMS_TRANSFORMATIONS_FEATUREFINDER_FEATUREFINDERIDENTIFICATIONALGORITHM_H
52 class IsotopeDistribution;
67 std::vector<PeptideIdentification> peptides,
68 const std::vector<ProteinIdentification>& proteins,
69 std::vector<PeptideIdentification> peptides_ext,
70 std::vector<ProteinIdentification> proteins_ext,
94 typedef std::multimap<double, PeptideIdentification*>
RTMap;
96 typedef std::map<Int, std::pair<RTMap, RTMap> >
ChargeMap;
138 void updateMembers_()
override;
154 } feature_filter_quality_;
163 } feature_filter_peptides_;
171 const String& seq1 = p1.
getHits()[0].getSequence().toString();
172 const String& seq2 = p2.
getHits()[0].getSequence().toString();
177 if (charge1 == charge2)
181 return charge1 < charge2;
220 void generateTransitions_(
const String& peptide_id,
double mz,
Int charge,
226 void getRTRegions_(
ChargeMap& peptide_data, std::vector<RTRegion>& rt_regions)
const;
228 void annotateFeaturesFinalizeAssay_(
230 std::map<
Size, std::vector<PeptideIdentification*> >& feat_ids,
236 void ensureConvexHulls_(
Feature& feature);
238 void postProcess_(
FeatureMap& features,
bool with_external_ids);
241 void statistics_(
const FeatureMap& features)
const;
245 void createAssayLibrary_(
const PeptideMap::iterator& begin,
const PeptideMap::iterator& end,
PeptideRefRTMap& ref_rt_map);
249 bool external =
false)
const;
251 void checkNumObservations_(
Size n_pos,
Size n_neg,
const String& note =
"")
const;
253 void getUnbiasedSample_(
const std::multimap<
double, std::pair<Size, bool> >& valid_obs,
254 std::map<Size, Int>& training_labels);
256 void getRandomSample_(std::map<Size, Int>& training_labels);
260 void filterFeaturesFinalizeAssay_(
Feature& best_feature,
double best_quality,
261 const double quality_cutoff);
263 void filterFeatures_(
FeatureMap& features,
bool classified);
269 template <
typename It>
270 std::vector<std::pair<It,It>>
271 chunk_(It range_from, It range_to,
const std::ptrdiff_t batch_size)
276 using std::make_pair;
278 using diff_t = std::ptrdiff_t;
281 const diff_t total {distance(range_from, range_to)};
282 const diff_t num {total / batch_size};
284 vector<pair<It,It>> chunks(num);
286 It batch_end {range_from};
289 std::generate(begin(chunks), end(chunks), [&batch_end, batch_size]()
291 It batch_start {batch_end };
293 std::advance(batch_end, batch_size);
294 return make_pair(batch_start, batch_end);
300 chunks.emplace_back(range_from, range_to);
304 chunks.back().second = range_to;
bool operator()(const Feature &f1, const Feature &f2)
Definition: FeatureFinderIdentificationAlgorithm.h:190
Size batch_size_
nr of peptides to use at the same time during chromatogram extraction
Definition: FeatureFinderIdentificationAlgorithm.h:107
Base class for TOPP applications.
Definition: TOPPBase.h:144
std::map< AASequence, ChargeMap > PeptideMap
mapping: sequence -> charge -> internal/external ID information
Definition: FeatureFinderIdentificationAlgorithm.h:98
bool operator()(const Feature &feature)
Definition: FeatureFinderIdentificationAlgorithm.h:159
void store(const String &filename, const TargetedExperiment &id) const
Stores a map in a TraML file.
PeakMap ms_data_
input LC-MS data
Definition: FeatureFinderIdentificationAlgorithm.h:202
MRMFeatureFinderScoring feat_finder_
OpenSWATH feature finder.
Definition: FeatureFinderIdentificationAlgorithm.h:215
PeakMap & getChromatograms()
Definition: FeatureFinderIdentificationAlgorithm.h:80
Size svm_n_parts_
number of partitions for SVM cross-validation
Definition: FeatureFinderIdentificationAlgorithm.h:130
const TargetedExperiment & getLibrary() const
Definition: FeatureFinderIdentificationAlgorithm.h:87
double isotope_pmin_
min. isotope probability for peptide assay
Definition: FeatureFinderIdentificationAlgorithm.h:114
double start
Definition: FeatureFinderIdentificationAlgorithm.h:143
PeptideMap peptide_map_
Definition: FeatureFinderIdentificationAlgorithm.h:102
PeakMap & getMSData()
Definition: FeatureFinderIdentificationAlgorithm.h:77
Represents a peptide (amino acid sequence)
Definition: TargetedExperimentHelper.h:370
ChargeMap ids
internal/external peptide IDs (per charge) in this region
Definition: FeatureFinderIdentificationAlgorithm.h:144
The MRMFeatureFinder finds and scores peaks of transitions that co-elute.
Definition: MRMFeatureFinderScoring.h:93
QualityType getOverallQuality() const
Non-mutable access to the overall quality.
void store(const String &filename, const PeakMap &map) const
Stores a map in an MzML file.
double peak_width_
Definition: FeatureFinderIdentificationAlgorithm.h:119
ProgressLogger & getProgressLogger()
Definition: FeatureFinderIdentificationAlgorithm.h:83
File adapter for MzML files.
Definition: MzMLFile.h:55
A more convenient string class.
Definition: String.h:58
In-Memory representation of a mass spectrometry experiment.
Definition: MSExperiment.h:77
std::map< String, double > isotope_probs_
isotope probabilities of transitions
Definition: FeatureFinderIdentificationAlgorithm.h:214
Helper struct for mass traces used in FeatureFinderAlgorithmPicked.
Definition: FeatureFinderAlgorithmPickedHelperStructs.h:79
comparison functor for features
Definition: FeatureFinderIdentificationAlgorithm.h:188
void addMSLevel(int level)
adds a desired MS level for peaks to load
size_t Size
Size type e.g. used as variable which can hold result of size()
Definition: Types.h:127
const ProgressLogger & getProgressLogger() const
Definition: FeatureFinderIdentificationAlgorithm.h:84
double signal_to_noise_
Definition: FeatureFinderIdentificationAlgorithm.h:121
bool operator()(const PeptideIdentification &p1, const PeptideIdentification &p2)
Definition: FeatureFinderIdentificationAlgorithm.h:168
void setPrimaryMSRunPath(const StringList &s)
set the file path to the primary MS run (usually the mzML file obtained after data conversion from ra...
void run(std::vector< PeptideIdentification > peptides, const std::vector< ProteinIdentification > &proteins, std::vector< PeptideIdentification > peptides_ext, std::vector< ProteinIdentification > proteins_ext, FeatureMap &features, const FeatureMap &seeds=FeatureMap())
String candidates_out_
Definition: FeatureFinderIdentificationAlgorithm.h:134
const std::vector< PeptideIdentification > & getPeptideIdentifications() const
returns a const reference to the PeptideIdentification vector
std::multiset< double > svm_probs_external_
SVM probabilities for "external" features (for FDR calculation):
Definition: FeatureFinderIdentificationAlgorithm.h:209
TargetedExperiment library_
accumulated assays for peptides
Definition: FeatureFinderIdentificationAlgorithm.h:204
FeatureFinderAlgorithmPickedHelperStructs::MassTrace MassTrace
Definition: FeatureFinderIdentificationAlgorithm.h:90
TargetedExperiment & getLibrary()
Definition: FeatureFinderIdentificationAlgorithm.h:86
A base class for all classes handling default parameters.
Definition: DefaultParamHandler.h:91
const std::vector< PeptideHit > & getHits() const
returns the peptide hits as const
Definition: FeatureFinderIdentificationAlgorithm.h:56
Size n_external_peps_
number of external peptides
Definition: FeatureFinderIdentificationAlgorithm.h:105
predicate for filtering features by overall quality:
Definition: FeatureFinderIdentificationAlgorithm.h:148
void load(const String &filename, std::vector< ProteinIdentification > &protein_ids, std::vector< PeptideIdentification > &peptide_ids)
Loads the identifications of an idXML file without identifier.
Size svm_n_samples_
number of samples for SVM training
Definition: FeatureFinderIdentificationAlgorithm.h:131
Main OpenMS namespace.
Definition: FeatureDeconvolution.h:46
std::map< Int, std::pair< RTMap, RTMap > > ChargeMap
mapping: charge -> internal/external: (RT -> pointer to peptide)
Definition: FeatureFinderIdentificationAlgorithm.h:96
File adapter for HUPO PSI TraML files.
Definition: TraMLFile.h:63
Base class for all classes that want to report their progress.
Definition: ProgressLogger.h:54
predicate for filtering features by assigned peptides:
Definition: FeatureFinderIdentificationAlgorithm.h:157
double svm_min_prob_
Definition: FeatureFinderIdentificationAlgorithm.h:126
void load(const String &filename, PeakMap &map)
Loads a map from a MzML file. Spectra and chromatograms are sorted by default (this can be disabled u...
double mz_window_
m/z window width
Definition: FeatureFinderIdentificationAlgorithm.h:109
double svm_quality_cutoff
Definition: FeatureFinderIdentificationAlgorithm.h:129
void load(const String &filename, FeatureMap &feature_map)
loads the file with name filename into map and calls updateRanges().
Size debug_level_
Definition: FeatureFinderIdentificationAlgorithm.h:136
std::map< String, std::pair< RTMap, RTMap > > PeptideRefRTMap
mapping: peptide ref. -> int./ext.: (RT -> pointer to peptide)
Definition: FeatureFinderIdentificationAlgorithm.h:100
double mapping_tolerance_
RT tolerance for mapping IDs to features.
Definition: FeatureFinderIdentificationAlgorithm.h:112
void setParameters(const Param ¶m)
Sets the parameters.
StringList svm_predictor_names_
Definition: FeatureFinderIdentificationAlgorithm.h:127
const Param & getDefaults() const
Non-mutable access to the default parameters.
bool operator()(const Feature &feature)
Definition: FeatureFinderIdentificationAlgorithm.h:150
std::vector< String > StringList
Vector of String.
Definition: ListUtils.h:70
void store(const String &filename, const FeatureMap &feature_map)
stores the map feature_map in file with name filename.
CoordinateType getRT() const
Returns the RT coordinate (index 0)
Definition: Peak2D.h:208
const PeakMap & getChromatograms() const
Definition: FeatureFinderIdentificationAlgorithm.h:81
int main(int argc, const char **argv)
Definition: INIFileEditor.cpp:73
region in RT in which a peptide elutes:
Definition: FeatureFinderIdentificationAlgorithm.h:141
std::multimap< double, PeptideIdentification * > RTMap
mapping: RT (not necessarily unique) -> pointer to peptide
Definition: FeatureFinderIdentificationAlgorithm.h:94
A container for features.
Definition: FeatureMap.h:95
Definition: IsotopeDistribution.h:64
Represents the peptide hits for a spectrum.
Definition: PeptideIdentification.h:62
void remove(const String &key)
Remove the entry key or a section key (when suffix is ':')
An LC-MS feature.
Definition: Feature.h:70
Size n_isotopes_
number of isotopes for peptide assay
Definition: FeatureFinderIdentificationAlgorithm.h:115
comparison functor for (unassigned) peptide IDs
Definition: FeatureFinderIdentificationAlgorithm.h:166
Size n_internal_features_
internal feature counter (for FDR calculation)
Definition: FeatureFinderIdentificationAlgorithm.h:210
double rt_quantile_
Definition: FeatureFinderIdentificationAlgorithm.h:117
This class provides Input/Output functionality for feature maps.
Definition: FeatureXMLFile.h:68
Quantitation.
Definition: DataProcessing.h:72
FeatureFinderAlgorithmPickedHelperStructs::MassTraces MassTraces
Definition: FeatureFinderIdentificationAlgorithm.h:91
const PeakMap & getMSData() const
Definition: FeatureFinderIdentificationAlgorithm.h:78
Management and storage of parameters / INI files.
Definition: Param.h:73
double min_peak_width_
Definition: FeatureFinderIdentificationAlgorithm.h:120
void clear(bool clear_meta_data)
Clears all data and meta data.
double rt_window_
RT window width.
Definition: FeatureFinderIdentificationAlgorithm.h:108
Helper struct for a collection of mass traces used in FeatureFinderAlgorithmPicked.
Definition: FeatureFinderAlgorithmPickedHelperStructs.h:109
A description of a targeted experiment containing precursor and production ions.
Definition: TargetedExperiment.h:64
#define OPENMS_LOG_INFO
Macro if a information, e.g. a status should be reported.
Definition: LogStream.h:465
bool mz_window_ppm_
m/z window width is given in PPM (not Da)?
Definition: FeatureFinderIdentificationAlgorithm.h:110
Size n_external_features_
Definition: FeatureFinderIdentificationAlgorithm.h:211
std::map< double, std::pair< Size, Size > > svm_probs_internal_
SVM probability -> number of pos./neg. features (for FDR calculation):
Definition: FeatureFinderIdentificationAlgorithm.h:207
PeakMap chrom_data_
accumulated chromatograms (XICs)
Definition: FeatureFinderIdentificationAlgorithm.h:203
void runOnCandidates(FeatureMap &features)
Data filtering or extraction.
Definition: DataProcessing.h:71
void insert(const String &prefix, const Param ¶m)
Size n_internal_peps_
number of internal peptide
Definition: FeatureFinderIdentificationAlgorithm.h:104
PeakFileOptions & getOptions()
Mutable access to the options for loading/storing.
void setLogType(LogType type) const
Sets the progress log that should be used. The default type is NONE!
ProgressLogger prog_log_
Definition: FeatureFinderIdentificationAlgorithm.h:217
String elution_model_
Definition: FeatureFinderIdentificationAlgorithm.h:123
String svm_xval_out_
Definition: FeatureFinderIdentificationAlgorithm.h:128
Used to load and store idXML files.
Definition: IdXMLFile.h:63
double getRT() const
returns the RT of the MS2 spectrum where the identification occurred
std::vector< std::pair< It, It > > chunk_(It range_from, It range_to, const std::ptrdiff_t batch_size)
Definition: FeatureFinderIdentificationAlgorithm.h:271