OpenMS
IDRipper.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 // $Maintainer: Timo Sachsenberg$
6 // $Authors: Immanuel Luhn, Leon Kuchenbecker$
7 // --------------------------------------------------------------------------
8 #pragma once
9 
13 #include <unordered_map>
14 
15 
16 namespace OpenMS
17 {
27  class OPENMS_DLLAPI IDRipper :
28  public DefaultParamHandler
29  {
30 public:
32  enum OriginAnnotationFormat { FILE_ORIGIN = 0, MAP_INDEX = 1, ID_MERGE_INDEX = 2, UNKNOWN_OAF = 3, SIZE_OF_ORIGIN_ANNOTATION_FORMAT = 4 };
33 
35  static const std::array<std::string, SIZE_OF_ORIGIN_ANNOTATION_FORMAT> names_of_OriginAnnotationFormat;
36 
38  struct OPENMS_DLLAPI IdentificationRuns
39  {
41  std::map<String, UInt> index_map;
43  std::vector<StringList> spectra_data;
44 
46  IdentificationRuns(const std::vector<ProteinIdentification>& prot_ids);
47  };
48 
50  struct OPENMS_DLLAPI RipFileIdentifier
51  {
53  UInt ident_run_idx{};
55  UInt file_origin_idx{};
60 
63  const PeptideIdentification& pep_id,
64  const std::map<String, UInt>& file_origin_map,
65  const IDRipper::OriginAnnotationFormat origin_annotation_fmt,
66  bool split_ident_runs);
67 
70 
73 
75  const String & getOriginFullname() const;
76 
78  const String & getOutputBasename() const;
79  };
80 
83  {
84  bool operator()(const RipFileIdentifier& left, const RipFileIdentifier& right) const;
85  };
86 
88  struct OPENMS_DLLAPI RipFileContent
89  {
91  std::vector<ProteinIdentification> prot_idents;
93  std::vector<PeptideIdentification> pep_idents;
95  RipFileContent(const std::vector<ProteinIdentification>& prot_idents, const std::vector<PeptideIdentification>& pep_idents)
96  : prot_idents(prot_idents), pep_idents(pep_idents) {}
98  const std::vector<ProteinIdentification> & getProteinIdentifications();
100  const std::vector<PeptideIdentification> & getPeptideIdentifications();
101  };
102 
104  typedef std::map<RipFileIdentifier, RipFileContent, RipFileIdentifierIdxComparator> RipFileMap;
105 
108 
110  ~IDRipper() override;
111 
124  void rip(
125  RipFileMap& ripped,
126  std::vector<ProteinIdentification>& proteins,
127  std::vector<PeptideIdentification>& peptides,
128  bool numeric_filenames,
129  bool split_ident_runs);
130 
143  // Autowrap compatible wrapper for rip(RipFileMap,...)
144  void rip(
145  std::vector<RipFileIdentifier> & rfis,
146  std::vector<RipFileContent> & rfcs,
147  std::vector<ProteinIdentification>& proteins,
148  std::vector<PeptideIdentification>& peptides,
149  bool numeric_filenames,
150  bool split_ident_runs);
151 
152 private:
153  // Not implemented
155  IDRipper(const IDRipper & rhs);
156 
157  // Not implemented
159  IDRipper & operator=(const IDRipper & rhs);
160 
162  OriginAnnotationFormat detectOriginAnnotationFormat_(std::map<String, UInt> & file_origin_map, const std::vector<PeptideIdentification> & peptide_idents);
164  void getProteinHits_(std::vector<ProteinHit> & result, const std::unordered_map<String, const ProteinHit*> & acc2protein_hits, const std::set<String> & protein_accessions);
166  std::set<String> getProteinAccessions_(const std::vector<PeptideHit> & peptide_hits);
170  bool registerBasename_(std::map<String, std::pair<UInt, UInt> >& basename_to_numeric, const IDRipper::RipFileIdentifier& rfi);
172  bool setOriginAnnotationMode_(short& mode, short const new_value);
173  };
174 
175 } // namespace OpenMS
A base class for all classes handling default parameters.
Definition: DefaultParamHandler.h:66
Ripping protein/peptide identification according their file origin.
Definition: IDRipper.h:29
bool registerBasename_(std::map< String, std::pair< UInt, UInt > > &basename_to_numeric, const IDRipper::RipFileIdentifier &rfi)
helper function, register a potential output file basename to detect duplicate output basenames
void rip(RipFileMap &ripped, std::vector< ProteinIdentification > &proteins, std::vector< PeptideIdentification > &peptides, bool numeric_filenames, bool split_ident_runs)
Ripping protein/peptide identification according their file origin.
IDRipper()
Default constructor.
~IDRipper() override
Destructor.
static const std::array< std::string, SIZE_OF_ORIGIN_ANNOTATION_FORMAT > names_of_OriginAnnotationFormat
String representations for the OriginAnnotationFormat enum.
Definition: IDRipper.h:35
void getProteinHits_(std::vector< ProteinHit > &result, const std::unordered_map< String, const ProteinHit * > &acc2protein_hits, const std::set< String > &protein_accessions)
helper function, extracts all protein hits that match the protein accession
OriginAnnotationFormat detectOriginAnnotationFormat_(std::map< String, UInt > &file_origin_map, const std::vector< PeptideIdentification > &peptide_idents)
helper function, detects file origin annotation standard from collections of protein and peptide hits
std::map< RipFileIdentifier, RipFileContent, RipFileIdentifierIdxComparator > RipFileMap
Represents the result of an IDRipper process, a map assigning file content to output file identifiers...
Definition: IDRipper.h:104
int getProteinIdentification_(const PeptideIdentification &pep_ident, const IdentificationRuns &id_runs)
helper function, returns the index of the protein identification for the given peptide identification...
OriginAnnotationFormat
Possible input file encodings for the origin as used by different versions of IDMerger.
Definition: IDRipper.h:32
IDRipper(const IDRipper &rhs)
Copy constructor.
bool setOriginAnnotationMode_(short &mode, short const new_value)
helper function, sets the value of mode to new_value and returns true if the old value was identical ...
void rip(std::vector< RipFileIdentifier > &rfis, std::vector< RipFileContent > &rfcs, std::vector< ProteinIdentification > &proteins, std::vector< PeptideIdentification > &peptides, bool numeric_filenames, bool split_ident_runs)
Ripping protein/peptide identification according their file origin.
IDRipper & operator=(const IDRipper &rhs)
Assignment.
std::set< String > getProteinAccessions_(const std::vector< PeptideHit > &peptide_hits)
helper function, returns the string representation of the peptide hit accession
Represents the peptide hits for a spectrum.
Definition: PeptideIdentification.h:39
A more convenient string class.
Definition: String.h:34
unsigned int UInt
Unsigned integer type.
Definition: Types.h:68
const std::string ID_MERGE_INDEX
Definition: Constants.h:297
Main OpenMS namespace.
Definition: FeatureDeconvolution.h:22
Represents a set of IdentificationRuns.
Definition: IDRipper.h:39
IdentificationRuns(const std::vector< ProteinIdentification > &prot_ids)
Generates a new IdentificationRuns object from a vector of ProteinIdentification objects.
std::vector< StringList > spectra_data
Maps the list of spectra data elements to every IdentificationRun index.
Definition: IDRipper.h:43
std::map< String, UInt > index_map
Maps a unique index to every IdentificationRun string representation (getIdentifier()).
Definition: IDRipper.h:41
Represents the content of an IDRipper output file.
Definition: IDRipper.h:89
const std::vector< ProteinIdentification > & getProteinIdentifications()
Get protein identifications.
std::vector< PeptideIdentification > pep_idents
Peptide identifications.
Definition: IDRipper.h:93
RipFileContent(const std::vector< ProteinIdentification > &prot_idents, const std::vector< PeptideIdentification > &pep_idents)
Constructs a new RipFileContent object.
Definition: IDRipper.h:95
const std::vector< PeptideIdentification > & getPeptideIdentifications()
Get peptide identifications.
std::vector< ProteinIdentification > prot_idents
Protein identifications.
Definition: IDRipper.h:91
Provides a 'less' operation for RipFileIdentifiers that ignores the out_basename and origin_fullname ...
Definition: IDRipper.h:83
bool operator()(const RipFileIdentifier &left, const RipFileIdentifier &right) const
Identifies an IDRipper output file.
Definition: IDRipper.h:51
RipFileIdentifier(const IDRipper::IdentificationRuns &id_runs, const PeptideIdentification &pep_id, const std::map< String, UInt > &file_origin_map, const IDRipper::OriginAnnotationFormat origin_annotation_fmt, bool split_ident_runs)
Constructs a new RipFileIdentifier object.
String out_basename
The output basename derived from the file_origin / spectra_data element.
Definition: IDRipper.h:57
const String & getOutputBasename() const
Get output base name.
const String & getOriginFullname() const
Get origin full name.
UInt getFileOriginIdx() const
Get file origin index.
String origin_fullname
The full length origin read from the file_origin / spectra_data element.
Definition: IDRipper.h:59
UInt getIdentRunIdx() const
Get identification run index.