OpenMS
AScore.h
Go to the documentation of this file.
1 // Copyright (c) 2002-2023, The OpenMS Team -- EKU Tuebingen, ETH Zurich, and FU Berlin
2 // SPDX-License-Identifier: BSD-3-Clause
3 //
4 // --------------------------------------------------------------------------
5 // $Maintainer: Petra Gutenbrunner $
6 // $Authors: David Wojnar, Timo Sachsenberg, Petra Gutenbrunner $
7 // --------------------------------------------------------------------------
8 
9 #pragma once
10 
11 #include <OpenMS/CONCEPT/Types.h>
17 
18 #include <limits>
19 #include <vector>
20 
21 namespace OpenMS
22 {
23  class PeptideHit;
24  class AASequence;
25 
27  {
28  public:
29 
36  };
37 
46  class OPENMS_DLLAPI AScore: public DefaultParamHandler
47  {
48  friend struct PScore;
49 
50  public:
52  AScore();
53 
55  ~AScore() override;
56 
65  PeptideHit compute(const PeptideHit& hit, PeakSpectrum& real_spectrum);
66 
67  protected:
68  int compareMZ_(double mz1, double mz2) const;
69 
73  template <class InputIterator1, class InputIterator2, class OutputIterator>
74  OutputIterator getSpectrumDifference_(InputIterator1 first1, InputIterator1 last1,
75  InputIterator2 first2, InputIterator2 last2, OutputIterator result) const
76  {
77  while (first1 != last1 && first2 != last2)
78  {
79  double mz1 = first1->getMZ();
80  double mz2 = first2->getMZ();
81  int val = compareMZ_(mz1, mz2);
82 
83  if (val == -1)
84  {
85  *result = *first1;
86  ++result;
87  ++first1;
88  }
89  else if (val == 1)
90  {
91  ++first2;
92  }
93  else // check if more ions are within the same tolerance. If so, these can not be site determining ions
94  {
95  //check mz2 until no match
96  ++first2;
97  if (first2 != last2)
98  {
99  int ret = compareMZ_(mz1, first2->getMZ());
100  while (ret == 0 && first2 != last2)
101  {
102  ++first2;
103  ret = compareMZ_(mz1, first2->getMZ());
104  }
105  }
106 
107  //check mz1 until no match
108  ++first1;
109  if (first1 != last1)
110  {
111  int ret = compareMZ_(first1->getMZ(), mz2);
112  while (ret == 0 && first1 != last1)
113  {
114  ++first1;
115  ret = compareMZ_(first1->getMZ(), mz2);
116  }
117  }
118  }
119  }
120  return std::copy(first1, last1, result);
121  }
122 
124  void computeSiteDeterminingIons_(const std::vector<PeakSpectrum>& th_spectra, const ProbablePhosphoSites& candidates, std::vector<PeakSpectrum>& site_determining_ions) const;
125 
127  std::vector<Size> getSites_(const AASequence& without_phospho) const;
128 
130  std::vector<std::vector<Size>> computePermutations_(const std::vector<Size>& sites, Int n_phosphorylation_events) const;
131 
133  Size numberOfMatchedIons_(const PeakSpectrum& th, const PeakSpectrum& windows, Size depth) const;
134 
136  double peptideScore_(const std::vector<double>& scores) const;
137 
142  void determineHighestScoringPermutations_(const std::vector<std::vector<double>>& peptide_site_scores, std::vector<ProbablePhosphoSites>& sites, const std::vector<std::vector<Size>>& permutations, std::multimap<double, Size>& ranking) const;
143 
145  double computeBaseProbability_(double ppm_reference_mz) const;
146 
148  double computeCumulativeScore_(Size N, Size n, double p) const;
149 
151  Size numberOfPhosphoEvents_(const String& sequence) const;
152 
155 
157  std::vector<PeakSpectrum> createTheoreticalSpectra_(const std::vector<std::vector<Size>>& permutations, const AASequence& seq_without_phospho) const;
158 
160  std::vector<PeakSpectrum> peakPickingPerWindowsInSpectrum_(PeakSpectrum& real_spectrum) const;
161 
163  std::vector<std::vector<double>> calculatePermutationPeptideScores_(std::vector<PeakSpectrum>& th_spectra, const std::vector<PeakSpectrum>& windows_top10) const;
164 
166  std::multimap<double, Size> rankWeightedPermutationPeptideScores_(const std::vector<std::vector<double>>& peptide_site_scores) const;
167 
169  void updateMembers_() override;
170 
171  // variables:
178  };
179 
180 } // namespace OpenMS
Representation of a peptide/protein sequence.
Definition: AASequence.h:86
Implementation of the Ascore For a given peptide sequence and its MS/MS spectrum it identifies the mo...
Definition: AScore.h:47
std::vector< PeakSpectrum > createTheoreticalSpectra_(const std::vector< std::vector< Size >> &permutations, const AASequence &seq_without_phospho) const
Create theoretical spectra with all combinations with the number of phosphorylation events.
std::vector< std::vector< Size > > computePermutations_(const std::vector< Size > &sites, Int n_phosphorylation_events) const
calculate all n_phosphorylation_events sized sets of phospho sites (all versions of the peptides with...
std::vector< std::vector< double > > calculatePermutationPeptideScores_(std::vector< PeakSpectrum > &th_spectra, const std::vector< PeakSpectrum > &windows_top10) const
Create 10 scores for each theoretical spectrum (permutation), according to Beausoleil et al....
AScore()
Default constructor.
double computeCumulativeScore_(Size N, Size n, double p) const
Computes the cumulative binomial probabilities.
double unambiguous_score_
Score for unambiguous assignments (all sites phosphorylated)
Definition: AScore.h:176
AASequence removePhosphositesFromSequence_(const String &sequence) const
Create variant of the peptide with all phosphorylations removed.
std::multimap< double, Size > rankWeightedPermutationPeptideScores_(const std::vector< std::vector< double >> &peptide_site_scores) const
Rank weighted permutation scores ascending.
Size numberOfPhosphoEvents_(const String &sequence) const
Computes number of phospho events in a sequence.
~AScore() override
Destructor.
double peptideScore_(const std::vector< double > &scores) const
Computes the peptide score according to Beausoleil et al. page 1291.
std::vector< Size > getSites_(const AASequence &without_phospho) const
return all phospho sites
OutputIterator getSpectrumDifference_(InputIterator1 first1, InputIterator1 last1, InputIterator2 first2, InputIterator2 last2, OutputIterator result) const
Definition: AScore.h:74
PeptideHit compute(const PeptideHit &hit, PeakSpectrum &real_spectrum)
Computes the AScore and returns all computed phospho-sites. The saved sequences contain only phospho ...
void computeSiteDeterminingIons_(const std::vector< PeakSpectrum > &th_spectra, const ProbablePhosphoSites &candidates, std::vector< PeakSpectrum > &site_determining_ions) const
Computes the site determining_ions for the given AS and sequences in candidates.
int compareMZ_(double mz1, double mz2) const
std::vector< PeakSpectrum > peakPickingPerWindowsInSpectrum_(PeakSpectrum &real_spectrum) const
Pick top 10 intensity peaks for each 100 Da windows.
void updateMembers_() override
Reimplemented from DefaultParamHandler.
double fragment_mass_tolerance_
Fragment mass tolerance for spectrum comparisons.
Definition: AScore.h:172
void determineHighestScoringPermutations_(const std::vector< std::vector< double >> &peptide_site_scores, std::vector< ProbablePhosphoSites > &sites, const std::vector< std::vector< Size >> &permutations, std::multimap< double, Size > &ranking) const
Finds the peptides with the highest PeptideScores and outputs all information for computing the AScor...
Size max_permutations_
Limit for number of sequence permutations that can be handled.
Definition: AScore.h:175
bool fragment_tolerance_ppm_
Is fragment mass tolerance given in ppm (or Da)?
Definition: AScore.h:173
double base_match_probability_
Probability of a match at a peak depth of 1.
Definition: AScore.h:177
Size numberOfMatchedIons_(const PeakSpectrum &th, const PeakSpectrum &windows, Size depth) const
Computes number of matched ions between windows and the given spectrum. All spectra have to be sorted...
Size max_peptide_length_
Limit for peptide lengths that can be analyzed.
Definition: AScore.h:174
double computeBaseProbability_(double ppm_reference_mz) const
Computes probability for a peak depth of one given spectra and mass_tolerance variables.
A base class for all classes handling default parameters.
Definition: DefaultParamHandler.h:66
The representation of a 1D spectrum.
Definition: MSSpectrum.h:44
Representation of a peptide hit.
Definition: PeptideHit.h:31
A more convenient string class.
Definition: String.h:34
int Int
Signed integer type.
Definition: Types.h:76
size_t Size
Size type e.g. used as variable which can hold result of size()
Definition: Types.h:101
Main OpenMS namespace.
Definition: FeatureDeconvolution.h:22
Size peak_depth
filtering level that gave rise to maximum discriminatory score
Definition: AScore.h:34
Size seq_2
index of permutation with site in unphosphorylated state
Definition: AScore.h:33
Size seq_1
index of best permutation with site in phosphorylated state
Definition: AScore.h:32
Size first
Definition: AScore.h:30
Size AScore
Definition: AScore.h:35
Size second
Definition: AScore.h:31
Definition: AScore.h:27
Implementation of the PScore PSM scoring algorithm.
Definition: PScore.h:24