OpenMS
FeatureFinderAlgorithmPicked.h
Go to the documentation of this file.
1 // Copyright (c) 2002-present, OpenMS Inc. -- EKU Tuebingen, ETH Zurich, and FU Berlin
2 // SPDX-License-Identifier: BSD-3-Clause
3 //
4 // --------------------------------------------------------------------------
5 // $Maintainer: Timo Sachsenberg $
6 // $Authors: Marc Sturm $
7 // --------------------------------------------------------------------------
8 
9 #pragma once
10 
18 
19 #include <fstream>
20 
21 namespace OpenMS
22 {
24  struct OPENMS_DLLAPI FeatureFinderDefs
25  {
28 
31 
34 
36  enum Flag {UNUSED, USED};
37 
39  class OPENMS_DLLAPI NoSuccessor :
41  {
42 public:
43  NoSuccessor(const char * file, int line, const char * function, const IndexPair & index) :
44  BaseException(file, line, function, "NoSuccessor", String("there is no successor/predecessor for the given Index: ") + String(index.first) + "/" + String(index.second)),
45  index_(index)
46  {
48  }
49 
50  ~NoSuccessor() noexcept override = default;
51 
52 protected:
53  IndexPair index_; // index without successor/predecessor
54  };
55  };
56 
57 
71  class OPENMS_DLLAPI FeatureFinderAlgorithmPicked :
72  public DefaultParamHandler, public ProgressLogger
73  {
74 public:
76 
81 
82 protected:
83  typedef Peak1D PeakType;
89 
90 public:
93 
94  void setSeeds(const FeatureMap& seeds);
95 
96  void setData(const MSExperiment& map, FeatureMap& features);
97 
98  void run(PeakMap& input_map,
99  FeatureMap& features,
100  const Param& param,
101  const FeatureMap& seeds);
102 
103  virtual Param getDefaultParameters() const
104  {
105  return defaults_;
106  }
107 protected:
108  void run();
109 
112 
114 
116  mutable std::ofstream log_;
118  bool debug_;
120  std::map<String, UInt> aborts_;
122  std::map<Seed, String> abort_reasons_;
125 
127 
132  double slope_bound_;
140  double min_rt_span_;
141  double max_rt_span_;
145 
147 
148  double intensity_rt_step_;
153  std::vector<std::vector<std::vector<double> > > intensity_thresholds_;
155 
157  std::vector<TheoreticalIsotopePattern> isotope_distributions_;
158 
159  // Docu in base class
160  void updateMembers_() override;
161 
163  void abort_(const Seed& seed, const String& reason);
164 
169  double intersection_(const Feature& f1, const Feature& f2) const;
170 
173 
181  double findBestIsotopeFit_(const Seed& center, UInt charge, IsotopePattern& best_pattern) const;
182 
190  void extendMassTraces_(const IsotopePattern& pattern, MassTraces& traces, Size meta_index_overall) const;
191 
210  void extendMassTrace_(MassTrace& trace, SignedSize spectrum_index, double mz, bool increase_rt, Size meta_index_overall, double min_rt = 0.0, double max_rt = 0.0) const;
211 
213  Size nearest_(double pos, const MSSpectrum& spec, Size start) const;
214 
224  void findIsotope_(double pos, Size spectrum_index, IsotopePattern& pattern, Size pattern_index, Size& peak_index) const;
225 
227  double positionScore_(double pos1, double pos2, double allowed_deviation) const;
228 
230  double isotopeScore_(const TheoreticalIsotopePattern& isotopes, IsotopePattern& pattern, bool consider_mz_distances) const;
231 
242  double intensityScore_(Size spectrum, Size peak) const;
243 
250  std::unique_ptr<TraceFitter> chooseTraceFitter_(double& tau);
251 
252  double intensityScore_(Size rt_bin, Size mz_bin, double intensity) const;
253 
260 
269  void cropFeature_(const std::shared_ptr<TraceFitter>& fitter,
270  const MassTraces& traces,
271  MassTraces& new_traces);
272 
296  bool checkFeatureQuality_(const std::shared_ptr<TraceFitter>& fitter,
297  MassTraces& feature_traces,
298  const double& seed_mz, const double& min_feature_score,
299  String& error_msg, double& fit_score, double& correlation, double& final_score);
300 
314  void writeFeatureDebugInfo_(const std::shared_ptr<TraceFitter>& fitter,
315  const MassTraces& traces,
316  const MassTraces& new_traces,
317  bool feature_ok, const String& error_msg, const double final_score, const Int plot_nr, const PeakType& peak,
318  const String& path = "debug/features/");
319 
321 private:
322 
327  };
328 
329 } // namespace OpenMS
330 
A base class for all classes handling default parameters.
Definition: DefaultParamHandler.h:66
Exception base class.
Definition: Exception.h:63
static void setMessage(const std::string &message)
FeatureFinderAlgorithm for picked peaks.
Definition: FeatureFinderAlgorithmPicked.h:73
double positionScore_(double pos1, double pos2, double allowed_deviation) const
Calculates a score between 0 and 1 for the m/z deviation of two peaks.
double slope_bound_
Max slope of mass trace intensities.
Definition: FeatureFinderAlgorithmPicked.h:132
FeatureFinderAlgorithmPickedHelperStructs::MassTraces MassTraces
Definition: FeatureFinderAlgorithmPicked.h:86
MapType map_
editable copy of the map
Definition: FeatureFinderAlgorithmPicked.h:111
void findIsotope_(double pos, Size spectrum_index, IsotopePattern &pattern, Size pattern_index, Size &peak_index) const
Searches for an isotopic peak in the current spectrum and the adjacent spectra.
double min_isotope_fit_
Minimum isotope pattern fit for a feature.
Definition: FeatureFinderAlgorithmPicked.h:138
double intensityScore_(Size rt_bin, Size mz_bin, double intensity) const
double findBestIsotopeFit_(const Seed &center, UInt charge, IsotopePattern &best_pattern) const
Finds the best fitting position of the isotopic pattern estimate defined by center.
UInt intensity_bins_
Number of bins (in RT and MZ) for intensity significance estimation.
Definition: FeatureFinderAlgorithmPicked.h:137
double max_feature_intersection_
Maximum allowed feature intersection (if larger, that one of the feature is removed)
Definition: FeatureFinderAlgorithmPicked.h:142
std::ofstream log_
Output stream for log/debug info.
Definition: FeatureFinderAlgorithmPicked.h:116
void extendMassTrace_(MassTrace &trace, SignedSize spectrum_index, double mz, bool increase_rt, Size meta_index_overall, double min_rt=0.0, double max_rt=0.0) const
Extends a single mass trace in one RT direction.
std::map< Seed, String > abort_reasons_
Array of abort reasons.
Definition: FeatureFinderAlgorithmPicked.h:122
std::unique_ptr< TraceFitter > chooseTraceFitter_(double &tau)
Choose a the best trace fitter for the current mass traces based on the user parameter (symmetric,...
FeatureMap seeds_
User-specified seed list.
Definition: FeatureFinderAlgorithmPicked.h:124
void extendMassTraces_(const IsotopePattern &pattern, MassTraces &traces, Size meta_index_overall) const
double mass_window_width_
Width of the isotope pattern mass bins.
Definition: FeatureFinderAlgorithmPicked.h:136
const TheoreticalIsotopePattern & getIsotopeDistribution_(double mass) const
Returns the isotope distribution for a certain mass window.
void run(PeakMap &input_map, FeatureMap &features, const Param &param, const FeatureMap &seeds)
double intensity_mz_step_
m/z bin width
Definition: FeatureFinderAlgorithmPicked.h:151
FeatureFinderAlgorithmPickedHelperStructs::IsotopePattern IsotopePattern
Definition: FeatureFinderAlgorithmPicked.h:88
SpectrumType::FloatDataArrays FloatDataArrays
Definition: FeatureFinderAlgorithmPicked.h:79
std::vector< std::vector< std::vector< double > > > intensity_thresholds_
Precalculated intensity 20-quantiles (binned)
Definition: FeatureFinderAlgorithmPicked.h:153
UInt max_missing_trace_peaks_
Stores mass_trace:max_missing.
Definition: FeatureFinderAlgorithmPicked.h:131
double min_rt_span_
Minimum RT range that has to be left after the fit.
Definition: FeatureFinderAlgorithmPicked.h:140
void abort_(const Seed &seed, const String &reason)
Writes the abort reason to the log file and counts occurrences for each reason.
UInt min_spectra_
Number of spectra that have to show the same mass (for finding a mass trace)
Definition: FeatureFinderAlgorithmPicked.h:130
double intensity_percentage_
Isotope pattern intensity contribution of required peaks.
Definition: FeatureFinderAlgorithmPicked.h:133
double trace_tolerance_
Stores isotopic_pattern:mz_tolerance.
Definition: FeatureFinderAlgorithmPicked.h:129
double intensityScore_(Size spectrum, Size peak) const
Compute the intensity score for the peak peak in spectrum spectrum.
double intersection_(const Feature &f1, const Feature &f2) const
bool checkFeatureQuality_(const std::shared_ptr< TraceFitter > &fitter, MassTraces &feature_traces, const double &seed_mz, const double &min_feature_score, String &error_msg, double &fit_score, double &correlation, double &final_score)
Checks the feature based on different score thresholds and model constraints.
double isotopeScore_(const TheoreticalIsotopePattern &isotopes, IsotopePattern &pattern, bool consider_mz_distances) const
Calculates a score between 0 and 1 for the correlation between theoretical and found isotope pattern.
FeatureFinderAlgorithmPickedHelperStructs::MassTrace MassTrace
Definition: FeatureFinderAlgorithmPicked.h:85
double min_trace_score_
Minimum quality of a traces.
Definition: FeatureFinderAlgorithmPicked.h:139
double optional_fit_improvement_
Minimal improvement for leaving out optional isotope.
Definition: FeatureFinderAlgorithmPicked.h:135
FeatureFinderAlgorithmPicked & operator=(const FeatureFinderAlgorithmPicked &)
Not implemented.
double intensity_percentage_optional_
Isotope pattern intensity contribution of optional peaks.
Definition: FeatureFinderAlgorithmPicked.h:134
void cropFeature_(const std::shared_ptr< TraceFitter > &fitter, const MassTraces &traces, MassTraces &new_traces)
Creates new mass traces new_traces based on the fitting result and the original traces traces.
void updateMembers_() override
This method is used to update extra member variables at the end of the setParameters() method.
FeatureFinderAlgorithmPickedHelperStructs::Seed Seed
Definition: FeatureFinderAlgorithmPicked.h:84
double max_rt_span_
Maximum RT range the model is allowed to span.
Definition: FeatureFinderAlgorithmPicked.h:141
double pattern_tolerance_
Stores mass_trace:mz_tolerance.
Definition: FeatureFinderAlgorithmPicked.h:128
void setData(const MSExperiment &map, FeatureMap &features)
void writeFeatureDebugInfo_(const std::shared_ptr< TraceFitter > &fitter, const MassTraces &traces, const MassTraces &new_traces, bool feature_ok, const String &error_msg, const double final_score, const Int plot_nr, const PeakType &peak, const String &path="debug/features/")
Creates several files containing plots and viewable data of the fitted mass trace.
Peak1D PeakType
Definition: FeatureFinderAlgorithmPicked.h:83
virtual Param getDefaultParameters() const
Definition: FeatureFinderAlgorithmPicked.h:103
FeatureMap * features_
Definition: FeatureFinderAlgorithmPicked.h:113
FeatureFinderAlgorithmPickedHelperStructs::TheoreticalIsotopePattern TheoreticalIsotopePattern
Definition: FeatureFinderAlgorithmPicked.h:87
FeatureFinderAlgorithmPicked(const FeatureFinderAlgorithmPicked &)
Not implemented.
String reported_mz_
The mass type that is reported for features. 'maximum' returns the m/z value of the highest mass trac...
Definition: FeatureFinderAlgorithmPicked.h:143
MapType::SpectrumType SpectrumType
Definition: FeatureFinderAlgorithmPicked.h:78
FeatureFinderAlgorithmPicked()
default constructor
void setSeeds(const FeatureMap &seeds)
Size nearest_(double pos, const MSSpectrum &spec, Size start) const
Returns the index of the peak nearest to m/z pos in spectrum spec (linear search starting from index ...
MSExperiment MapType
Definition: FeatureFinderAlgorithmPicked.h:77
bool debug_
debug flag
Definition: FeatureFinderAlgorithmPicked.h:118
std::vector< TheoreticalIsotopePattern > isotope_distributions_
Vector of precalculated isotope distributions for several mass windows.
Definition: FeatureFinderAlgorithmPicked.h:157
std::map< String, UInt > aborts_
Array of abort reasons.
Definition: FeatureFinderAlgorithmPicked.h:120
Exception that is thrown if a method an invalid IndexPair is given.
Definition: FeatureFinderAlgorithmPicked.h:41
~NoSuccessor() noexcept override=default
NoSuccessor(const char *file, int line, const char *function, const IndexPair &index)
Definition: FeatureFinderAlgorithmPicked.h:43
A container for features.
Definition: FeatureMap.h:80
An LC-MS feature.
Definition: Feature.h:46
In-Memory representation of a mass spectrometry run.
Definition: MSExperiment.h:46
The representation of a 1D spectrum.
Definition: MSSpectrum.h:44
std::vector< FloatDataArray > FloatDataArrays
Definition: MSSpectrum.h:96
Management and storage of parameters / INI files.
Definition: Param.h:44
A 1-dimensional raw data point or peak.
Definition: Peak1D.h:28
Base class for all classes that want to report their progress.
Definition: ProgressLogger.h:27
A more convenient string class.
Definition: String.h:34
int Int
Signed integer type.
Definition: Types.h:72
unsigned int UInt
Unsigned integer type.
Definition: Types.h:64
ptrdiff_t SignedSize
Signed Size type e.g. used as pointer difference.
Definition: Types.h:104
size_t Size
Size type e.g. used as variable which can hold result of size()
Definition: Types.h:97
Main OpenMS namespace.
Definition: openswathalgo/include/OpenMS/OPENSWATHALGO/DATAACCESS/ISpectrumAccess.h:19
Helper structure for a found isotope pattern used in FeatureFinderAlgorithmPicked.
Definition: FeatureFinderAlgorithmPickedHelperStructs.h:168
Helper struct for mass traces used in FeatureFinderAlgorithmPicked.
Definition: FeatureFinderAlgorithmPickedHelperStructs.h:54
Helper struct for a collection of mass traces used in FeatureFinderAlgorithmPicked.
Definition: FeatureFinderAlgorithmPickedHelperStructs.h:85
Helper structure for seeds used in FeatureFinderAlgorithmPicked.
Definition: FeatureFinderAlgorithmPickedHelperStructs.h:37
Helper structure for a theoretical isotope pattern used in FeatureFinderAlgorithmPicked.
Definition: FeatureFinderAlgorithmPickedHelperStructs.h:148
The purpose of this struct is to provide definitions of classes and typedefs which are used throughou...
Definition: FeatureFinderAlgorithmPicked.h:25
Flag
Flags that indicate if a peak is already used in a feature.
Definition: FeatureFinderAlgorithmPicked.h:36
IsotopeCluster::IndexPair IndexPair
Index to peak consisting of two UInts (scan index / peak index)
Definition: FeatureFinderAlgorithmPicked.h:27
IsotopeCluster::ChargedIndexSet ChargedIndexSet
Index to peak consisting of two UInts (scan index / peak index) with charge information.
Definition: FeatureFinderAlgorithmPicked.h:30
IsotopeCluster::IndexSet IndexSet
A set of peak indices.
Definition: FeatureFinderAlgorithmPicked.h:33
index set with associated charge estimate
Definition: IsotopeCluster.h:28
std::pair< Size, Size > IndexPair
An index e.g. in an MSExperiment.
Definition: IsotopeCluster.h:21
std::set< IndexPair > IndexSet
A set of index pairs, usually referring to an MSExperiment.
Definition: IsotopeCluster.h:23