OpenMS
OPXLHelper.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: Eugen Netz $
6 // $Authors: Eugen Netz $
7 // --------------------------------------------------------------------------
8 
9 #pragma once
10 
17 #include <numeric>
18 
19 namespace OpenMS
20 {
24  class OPENMS_DLLAPI OPXLHelper
25  {
26  public:
27 
33  {
34  bool operator() (const PeptideIdentification& a, const PeptideIdentification& b) const
35  {
36  if (!a.getHits().empty() && !b.getHits().empty())
37  {
38  return a.getHits()[0].getScore() < b.getHits()[0].getScore();
39  }
40  else
41  {
42  return false;
43  }
44  }
45  bool operator() (const PeptideIdentification& a, const double& b) const
46  {
47  if (!a.getHits().empty())
48  {
49  return a.getHits()[0].getScore() < b;
50  }
51  else
52  {
53  return false;
54  }
55  }
56  bool operator() (const double& a, const PeptideIdentification& b) const
57  {
58  if (!b.getHits().empty())
59  {
60  return a < b.getHits()[0].getScore();
61  }
62  else
63  {
64  return false;
65  }
66  }
67  };
68 
86  static std::vector<OPXLDataStructs::XLPrecursor> enumerateCrossLinksAndMasses(const std::vector<OPXLDataStructs::AASeqWithMass>& peptides, double cross_link_mass_light, const DoubleList& cross_link_mass_mono_link, const StringList& cross_link_residue1, const StringList& cross_link_residue2, const std::vector< double >& spectrum_precursors, std::vector< int >& precursor_correction_positions, double precursor_mass_tolerance, bool precursor_mass_tolerance_unit_ppm);
87 
108  static std::vector<OPXLDataStructs::AASeqWithMass> digestDatabase(std::vector<FASTAFile::FASTAEntry> fasta_db,
109  const EnzymaticDigestion& digestor, Size min_peptide_length, const StringList& cross_link_residue1, const StringList& cross_link_residue2,
110  const ModifiedPeptideGenerator::MapToResidueType& fixed_modifications,
111  const ModifiedPeptideGenerator::MapToResidueType& variable_modifications,
112  Size max_variable_mods_per_peptide);
113 
129  static std::vector <OPXLDataStructs::ProteinProteinCrossLink> buildCandidates(const std::vector< OPXLDataStructs::XLPrecursor > & candidates,
130  const std::vector< int > & precursor_corrections,
131  const std::vector< int > & precursor_correction_positions,
132  const std::vector<OPXLDataStructs::AASeqWithMass> & peptide_masses,
133  const StringList & cross_link_residue1,
134  const StringList & cross_link_residue2,
135  double cross_link_mass,
136  const DoubleList & cross_link_mass_mono_link,
137  const std::vector< double >& spectrum_precursor_vector,
138  const std::vector< double >& allowed_error_vector,
139  const String& cross_link_name);
140 
153  static void buildFragmentAnnotations(std::vector<PeptideHit::PeakAnnotation> & frag_annotations, const std::vector< std::pair< Size, Size > > & matching, const PeakSpectrum & theoretical_spectrum, const PeakSpectrum & experiment_spectrum);
154 
165  static void buildPeptideIDs(std::vector<PeptideIdentification> & peptide_ids, const std::vector< OPXLDataStructs::CrossLinkSpectrumMatch > & top_csms_spectrum, std::vector< std::vector< OPXLDataStructs::CrossLinkSpectrumMatch > > & all_top_csms, Size all_top_csms_current_index, const PeakMap & spectra, Size scan_index, Size scan_index_heavy);
166 
171  static void addProteinPositionMetaValues(std::vector< PeptideIdentification > & peptide_ids);
172 
177  static void addXLTargetDecoyMV(std::vector< PeptideIdentification > & peptide_ids);
178 
183  static void addBetaAccessions(std::vector< PeptideIdentification > & peptide_ids);
184 
189  static void removeBetaPeptideHits(std::vector< PeptideIdentification > & peptide_ids);
190 
196 
201  static void computeDeltaScores(std::vector< PeptideIdentification >& peptide_ids);
202 
213  static std::vector< PeptideIdentification > combineTopRanksFromPairs(std::vector< PeptideIdentification > & peptide_ids, Size number_top_hits);
214 
234  static std::vector <OPXLDataStructs::ProteinProteinCrossLink> collectPrecursorCandidates(const IntList& precursor_correction_steps,
235  double precursor_mass,
236  double precursor_mass_tolerance,
237  bool precursor_mass_tolerance_unit_ppm,
238  const std::vector<OPXLDataStructs::AASeqWithMass>& filtered_peptide_masses,
239  double cross_link_mass,
240  const DoubleList& cross_link_mass_mono_link,
241  const StringList& cross_link_residue1,
242  const StringList& cross_link_residue2,
243  String cross_link_name,
244  bool use_sequence_tags = false,
245  const std::vector<std::string>& tags = std::vector<std::string>());
246 
254  static double computePrecursorError(const OPXLDataStructs::CrossLinkSpectrumMatch& csm, double precursor_mz, int precursor_charge);
255 
263  static void isoPeakMeans(OPXLDataStructs::CrossLinkSpectrumMatch& csm, DataArrays::IntegerDataArray& num_iso_peaks_array, std::vector< std::pair< Size, Size > >& matched_spec_linear_alpha, std::vector< std::pair< Size, Size > >& matched_spec_linear_beta, std::vector< std::pair< Size, Size > >& matched_spec_xlinks_alpha, std::vector< std::pair< Size, Size > >& matched_spec_xlinks_beta);
264 
271  static void filterPrecursorsByTags(std::vector <OPXLDataStructs::XLPrecursor>& candidates, std::vector< int >& precursor_correction_positions, const std::vector<std::string>& tags);
272  };
273 }
Integer data array class.
Definition: DataArrays.h:29
Class for the enzymatic digestion of sequences.
Definition: EnzymaticDigestion.h:38
In-Memory representation of a mass spectrometry run.
Definition: MSExperiment.h:46
The representation of a 1D spectrum.
Definition: MSSpectrum.h:44
Definition: ModifiedPeptideGenerator.h:31
The CrossLinkSpectrumMatch struct represents a PSM between a ProteinProteinCrossLink and a spectrum i...
Definition: OPXLDataStructs.h:87
The OPXLHelper class contains functions needed by OpenPepXL and OpenPepXLLF to reduce duplicated code...
Definition: OPXLHelper.h:25
static void filterPrecursorsByTags(std::vector< OPXLDataStructs::XLPrecursor > &candidates, std::vector< int > &precursor_correction_positions, const std::vector< std::string > &tags)
Filters the list of candidates for cases that include at least one of the tags in at least one of the...
static std::vector< PeptideIdentification > combineTopRanksFromPairs(std::vector< PeptideIdentification > &peptide_ids, Size number_top_hits)
combines all hits to spectrum pairs with the same light spectrum into one ranked list
static void addBetaAccessions(std::vector< PeptideIdentification > &peptide_ids)
adds accessions_beta MetaValue to alpha peptides for TOPPView visualization and CSV table output
static void computeDeltaScores(std::vector< PeptideIdentification > &peptide_ids)
sorts PeptideHits for each PeptideIdentification by score and adds the delta score as a MetaValue
static std::vector< OPXLDataStructs::ProteinProteinCrossLink > collectPrecursorCandidates(const IntList &precursor_correction_steps, double precursor_mass, double precursor_mass_tolerance, bool precursor_mass_tolerance_unit_ppm, const std::vector< OPXLDataStructs::AASeqWithMass > &filtered_peptide_masses, double cross_link_mass, const DoubleList &cross_link_mass_mono_link, const StringList &cross_link_residue1, const StringList &cross_link_residue2, String cross_link_name, bool use_sequence_tags=false, const std::vector< std::string > &tags=std::vector< std::string >())
Searches for cross-link candidates for a MS/MS spectrum.
static void buildFragmentAnnotations(std::vector< PeptideHit::PeakAnnotation > &frag_annotations, const std::vector< std::pair< Size, Size > > &matching, const PeakSpectrum &theoretical_spectrum, const PeakSpectrum &experiment_spectrum)
Fills up the given FragmentAnnotation vector with annotations from a theoretical spectrum.
static void addProteinPositionMetaValues(std::vector< PeptideIdentification > &peptide_ids)
adds MetaValues for cross-link positions to PeptideHits
static void buildPeptideIDs(std::vector< PeptideIdentification > &peptide_ids, const std::vector< OPXLDataStructs::CrossLinkSpectrumMatch > &top_csms_spectrum, std::vector< std::vector< OPXLDataStructs::CrossLinkSpectrumMatch > > &all_top_csms, Size all_top_csms_current_index, const PeakMap &spectra, Size scan_index, Size scan_index_heavy)
Builds PeptideIdentifications and PeptideHits.
static void addPercolatorFeatureList(ProteinIdentification &prot_id)
adds the list of features that percolator should use for OpenPepXL
static void isoPeakMeans(OPXLDataStructs::CrossLinkSpectrumMatch &csm, DataArrays::IntegerDataArray &num_iso_peaks_array, std::vector< std::pair< Size, Size > > &matched_spec_linear_alpha, std::vector< std::pair< Size, Size > > &matched_spec_linear_beta, std::vector< std::pair< Size, Size > > &matched_spec_xlinks_alpha, std::vector< std::pair< Size, Size > > &matched_spec_xlinks_beta)
Computes the mass error of a precursor mass to a hit.
static void addXLTargetDecoyMV(std::vector< PeptideIdentification > &peptide_ids)
adds xl_target_decoy MetaValue that combines alpha and beta target_decoy info
static void removeBetaPeptideHits(std::vector< PeptideIdentification > &peptide_ids)
removes beta peptides from cross-link IDs, since all info is already contained in the alpha peptide h...
static std::vector< OPXLDataStructs::AASeqWithMass > digestDatabase(std::vector< FASTAFile::FASTAEntry > fasta_db, const EnzymaticDigestion &digestor, Size min_peptide_length, const StringList &cross_link_residue1, const StringList &cross_link_residue2, const ModifiedPeptideGenerator::MapToResidueType &fixed_modifications, const ModifiedPeptideGenerator::MapToResidueType &variable_modifications, Size max_variable_mods_per_peptide)
Digests a database with the given EnzymaticDigestion settings and precomputes masses for all peptides...
static double computePrecursorError(const OPXLDataStructs::CrossLinkSpectrumMatch &csm, double precursor_mz, int precursor_charge)
Computes the mass error of a precursor mass to a hit.
static std::vector< OPXLDataStructs::ProteinProteinCrossLink > buildCandidates(const std::vector< OPXLDataStructs::XLPrecursor > &candidates, const std::vector< int > &precursor_corrections, const std::vector< int > &precursor_correction_positions, const std::vector< OPXLDataStructs::AASeqWithMass > &peptide_masses, const StringList &cross_link_residue1, const StringList &cross_link_residue2, double cross_link_mass, const DoubleList &cross_link_mass_mono_link, const std::vector< double > &spectrum_precursor_vector, const std::vector< double > &allowed_error_vector, const String &cross_link_name)
Builds specific cross-link candidates with all possible combinations of linked positions from peptide...
static std::vector< OPXLDataStructs::XLPrecursor > enumerateCrossLinksAndMasses(const std::vector< OPXLDataStructs::AASeqWithMass > &peptides, double cross_link_mass_light, const DoubleList &cross_link_mass_mono_link, const StringList &cross_link_residue1, const StringList &cross_link_residue2, const std::vector< double > &spectrum_precursors, std::vector< int > &precursor_correction_positions, double precursor_mass_tolerance, bool precursor_mass_tolerance_unit_ppm)
Enumerates precursor masses for all candidates in an XL-MS search.
Represents the peptide hits for a spectrum.
Definition: PeptideIdentification.h:39
const std::vector< PeptideHit > & getHits() const
returns the peptide hits as const
Representation of a protein identification run.
Definition: ProteinIdentification.h:50
A more convenient string class.
Definition: String.h:34
size_t Size
Size type e.g. used as variable which can hold result of size()
Definition: Types.h:101
std::vector< Int > IntList
Vector of signed integers.
Definition: ListUtils.h:29
std::vector< String > StringList
Vector of String.
Definition: ListUtils.h:44
std::vector< double > DoubleList
Vector of double precision real types.
Definition: ListUtils.h:36
Main OpenMS namespace.
Definition: FeatureDeconvolution.h:22
A comparator for PeptideIdentifications that compares the scores in the first PeptideHit.
Definition: OPXLHelper.h:33