OpenMS
MzIdentMLDOMHandler.h
Go to the documentation of this file.
1 // Copyright (c) 2002-present, The OpenMS Team -- EKU Tuebingen, ETH Zurich, and FU Berlin
2 // SPDX-License-Identifier: BSD-3-Clause
3 //
4 // --------------------------------------------------------------------------
5 // $Maintainer: Mathias Walzer $
6 // $Authors: Mathias Walzer$
7 // --------------------------------------------------------------------------
8 
9 #pragma once
10 
20 
21 #include <xercesc/dom/DOM.hpp>
22 #include <xercesc/dom/DOMDocument.hpp>
23 #include <xercesc/dom/DOMDocumentType.hpp>
24 #include <xercesc/dom/DOMElement.hpp>
25 #include <xercesc/dom/DOMImplementation.hpp>
26 #include <xercesc/dom/DOMImplementationLS.hpp>
27 #include <xercesc/dom/DOMNodeIterator.hpp>
28 #include <xercesc/dom/DOMNodeList.hpp>
29 #include <xercesc/dom/DOMText.hpp>
30 #include <xercesc/framework/LocalFileFormatTarget.hpp>
31 #include <xercesc/framework/psvi/XSValue.hpp>
32 #include <xercesc/parsers/XercesDOMParser.hpp>
33 #include <xercesc/util/OutOfMemoryException.hpp>
34 #include <xercesc/util/PlatformUtils.hpp>
35 #include <xercesc/util/XMLString.hpp>
36 #include <xercesc/util/XMLUni.hpp>
37 
38 #include <list>
39 #include <map>
40 #include <stdexcept>
41 #include <string>
42 #include <vector>
43 
44 // Error codes
45 //enum {
46 // ERROR_ARGS = 1,
47 // ERROR_XERCES_INIT,
48 // ERROR_PARSE,
49 // ERROR_EMPTY_DOCUMENT
50 //};
51 
52 namespace OpenMS
53 {
54  class ProgressLogger;
55 
56  namespace Internal
57  {
69  class OPENMS_DLLAPI MzIdentMLDOMHandler
70  {
71 public:
75  MzIdentMLDOMHandler(const std::vector<ProteinIdentification>& pro_id, const std::vector<PeptideIdentification>& pep_id, const String& version, const ProgressLogger& logger);
76 
78  MzIdentMLDOMHandler(std::vector<ProteinIdentification>& pro_id, std::vector<PeptideIdentification>& pep_id, const String& version, const ProgressLogger& logger);
79 
83 
85  void readMzIdentMLFile(const std::string& mzid_file);
87  void writeMzIdentMLFile(const std::string& mzid_file);
88 
89 protected:
92 
97 
99  std::vector<ProteinIdentification>* pro_id_ = nullptr;
101  std::vector<PeptideIdentification>* pep_id_ = nullptr;
102 
104  const std::vector<ProteinIdentification>* cpro_id_ = nullptr;
106  const std::vector<PeptideIdentification>* cpep_id_ = nullptr;
107 
110 
112  ControlledVocabulary::CVTerm getChildWithName_(const String& parent_accession, const String& name) const;
113 
116 
118  std::pair<CVTermList, std::map<String, DataValue> > parseParamGroup_(xercesc::DOMNodeList* paramGroup);
119  CVTerm parseCvParam_(xercesc::DOMElement* param);
120  std::pair<String, DataValue> parseUserParam_(xercesc::DOMElement* param);
121  void parseAnalysisSoftwareList_(xercesc::DOMNodeList* analysisSoftwareElements);
122  void parseDBSequenceElements_(xercesc::DOMNodeList* dbSequenceElements);
123  void parsePeptideElements_(xercesc::DOMNodeList* peptideElements);
124  //AASequence parsePeptideSiblings_(xercesc::DOMNodeList* peptideSiblings);
125  AASequence parsePeptideSiblings_(xercesc::DOMElement* peptide);
126  void parsePeptideEvidenceElements_(xercesc::DOMNodeList* peptideEvidenceElements);
127  void parseSpectrumIdentificationElements_(xercesc::DOMNodeList* spectrumIdentificationElements);
128  void parseSpectrumIdentificationProtocolElements_(xercesc::DOMNodeList* spectrumIdentificationProtocolElements);
129  void parseInputElements_(xercesc::DOMNodeList* inputElements);
130  void parseSpectrumIdentificationListElements_(xercesc::DOMNodeList* spectrumIdentificationListElements);
131  void parseSpectrumIdentificationItemSetXLMS(std::set<String>::const_iterator set_it, std::multimap<String, int> xl_val_map, xercesc::DOMElement* element_res, const String& spectrumID);
132  void parseSpectrumIdentificationItemElement_(xercesc::DOMElement* spectrumIdentificationItemElement, PeptideIdentification& spectrum_identification, String& spectrumIdentificationList_ref);
133  void parseProteinDetectionHypothesisElement_(xercesc::DOMElement* proteinDetectionHypothesisElement, ProteinIdentification& protein_identification);
134  void parseProteinAmbiguityGroupElement_(xercesc::DOMElement* proteinAmbiguityGroupElement, ProteinIdentification& protein_identification);
135  void parseProteinDetectionListElements_(xercesc::DOMNodeList* proteinDetectionListElements);
136  static ProteinIdentification::SearchParameters findSearchParameters_(std::pair<CVTermList, std::map<String, DataValue> > as_params);
138 
140  void buildCvList_(xercesc::DOMElement* cvElements);
141  void buildAnalysisSoftwareList_(xercesc::DOMElement* analysisSoftwareElements);
142  void buildSequenceCollection_(xercesc::DOMElement* sequenceCollectionElements);
143  void buildAnalysisCollection_(xercesc::DOMElement* analysisCollectionElements);
144  void buildAnalysisProtocolCollection_(xercesc::DOMElement* protocolElements);
145  void buildInputDataCollection_(xercesc::DOMElement* inputElements);
146  void buildEnclosedCV_(xercesc::DOMElement* parentElement, const String& encel, const String& acc, const String& name, const String& cvref);
147  void buildAnalysisDataCollection_(xercesc::DOMElement* analysisElements);
149 
150 
151 private:
155 
158  {
161  };
164  {
165  int start;
166  int stop;
167  char pre;
168  char post;
169  bool idec;
170  };
172  struct DBSequence
173  {
178  };
181  {
186  };
189  {
191  long double mass_delta;
195  };
198  {
202  std::map<String, DataValue> parameter_ups;
203 // std::vector<ModificationParam> modification_parameter;
205  long double precursor_tolerance;
206  long double fragment_tolerance;
208  std::map<String, DataValue> threshold_ups;
209  };
212  {
217  };
218 
222 
223  xercesc::XercesDOMParser mzid_parser_;
224 
225  std::unique_ptr<XMLHandler> xml_handler_ = nullptr;
226 
227  //from AnalysisSoftware
230  //mapping from AnalysisSoftware
231  std::map<String, AnalysisSoftware> as_map_;
232 
233  //mapping from DataCollection Inputs
234  std::map<String, String> sr_map_;
235  std::map<String, String> sd_map_;
236  std::map<String, DatabaseInput> db_map_;
237 
238  //mapping from SpectrumIdentification - SpectrumIdentification will be the new IdentificationRuns
239  std::map<String, SpectrumIdentification> si_map_;
240  std::map<String, size_t> si_pro_map_;
241 
242  //mapping from SpectrumIdentificationProtocol
243  std::map<String, SpectrumIdentificationProtocol> sp_map_;
244 
245  //mapping from SequenceCollection
246  std::map<String, AASequence> pep_map_;
247  std::map<String, PeptideEvidence> pe_ev_map_;
248  std::map<String, String> pv_db_map_;
249  std::multimap<String, String> p_pv_map_;
250  std::map<String, DBSequence> db_sq_map_;
251 
252  std::list<std::list<String> > hit_pev_;
253 
255  std::map<String, String> xl_id_donor_map_;
256  //std::map<String, String> xl_id_acceptor_map_; ///< mapping Peptide id -> crosslink acceptor value
257  std::map<String, String> xl_id_acceptor_map_;
258  std::map<String, SignedSize> xl_donor_pos_map_;
259  std::map<String, SignedSize> xl_acceptor_pos_map_;
260  std::map<String, double> xl_mass_map_;
261  std::map<String, String> xl_mod_map_;
262 
263  };
264  } // namespace Internal
265 } // namespace OpenMS
266 
Representation of a peptide/protein sequence.
Definition: AASequence.h:86
Representation of controlled vocabulary term list.
Definition: CVTermList.h:28
Representation of controlled vocabulary term.
Definition: CVTerm.h:27
Definition: ControlledVocabulary.h:29
DateTime Class.
Definition: DateTime.h:33
XML DOM handler for MzIdentMLFile.
Definition: MzIdentMLDOMHandler.h:70
ControlledVocabulary::CVTerm getChildWithName_(const String &parent_accession, const String &name) const
Looks up a child CV term of parent_accession with the name name. If no such term is found,...
void readMzIdentMLFile(const std::string &mzid_file)
Provides the functionality of reading a mzid with a handler object.
std::map< String, SpectrumIdentification > si_map_
mapping SpectrumIdentification id -> SpectrumIdentification (id refs)
Definition: MzIdentMLDOMHandler.h:239
void buildAnalysisSoftwareList_(xercesc::DOMElement *analysisSoftwareElements)
std::map< String, DatabaseInput > db_map_
mapping database id -> DatabaseInput
Definition: MzIdentMLDOMHandler.h:236
CVTermList parameter_cvs
Definition: MzIdentMLDOMHandler.h:201
String search_database_ref
Definition: MzIdentMLDOMHandler.h:183
bool xl_ms_search_
is true when reading a file containing Cross-Linking MS search results
Definition: MzIdentMLDOMHandler.h:254
void parseSpectrumIdentificationItemElement_(xercesc::DOMElement *spectrumIdentificationItemElement, PeptideIdentification &spectrum_identification, String &spectrumIdentificationList_ref)
DateTime date
Definition: MzIdentMLDOMHandler.h:216
std::map< String, double > xl_mass_map_
mapping Peptide id -> cross-link mass
Definition: MzIdentMLDOMHandler.h:260
std::map< String, String > sd_map_
mapping spectradata id -> spectradata location
Definition: MzIdentMLDOMHandler.h:235
CVTerm parseCvParam_(xercesc::DOMElement *param)
std::map< String, DBSequence > db_sq_map_
mapping DBSequence id -> Sequence
Definition: MzIdentMLDOMHandler.h:250
XMLCh * xml_root_tag_ptr_
Definition: MzIdentMLDOMHandler.h:219
int start
Definition: MzIdentMLDOMHandler.h:165
String sequence
Definition: MzIdentMLDOMHandler.h:174
std::map< String, String > sr_map_
mapping sourcefile id -> sourcefile location
Definition: MzIdentMLDOMHandler.h:234
void parseProteinDetectionListElements_(xercesc::DOMNodeList *proteinDetectionListElements)
long double precursor_tolerance
Definition: MzIdentMLDOMHandler.h:205
void parseInputElements_(xercesc::DOMNodeList *inputElements)
std::pair< String, DataValue > parseUserParam_(xercesc::DOMElement *param)
String location
Definition: MzIdentMLDOMHandler.h:214
void parseProteinDetectionHypothesisElement_(xercesc::DOMElement *proteinDetectionHypothesisElement, ProteinIdentification &protein_identification)
std::map< String, SignedSize > xl_acceptor_pos_map_
mapping acceptor value -> cross-link modification location
Definition: MzIdentMLDOMHandler.h:259
const ProgressLogger & logger_
Progress logger.
Definition: MzIdentMLDOMHandler.h:91
ControlledVocabulary cv_
Controlled vocabulary (psi-ms from OpenMS/share/OpenMS/CV/psi-ms.obo)
Definition: MzIdentMLDOMHandler.h:94
String search_engine_
Definition: MzIdentMLDOMHandler.h:228
void buildAnalysisCollection_(xercesc::DOMElement *analysisCollectionElements)
std::map< String, PeptideEvidence > pe_ev_map_
mapping PeptideEvidence id -> PeptideEvidence
Definition: MzIdentMLDOMHandler.h:247
void parseSpectrumIdentificationItemSetXLMS(std::set< String >::const_iterator set_it, std::multimap< String, int > xl_val_map, xercesc::DOMElement *element_res, const String &spectrumID)
long double mass_delta
Definition: MzIdentMLDOMHandler.h:191
CVTermList modification_param_cvs
Definition: MzIdentMLDOMHandler.h:193
String version
Definition: MzIdentMLDOMHandler.h:160
static ProteinIdentification::SearchParameters findSearchParameters_(std::pair< CVTermList, std::map< String, DataValue > > as_params)
XMLCh * xml_cvparam_tag_ptr_
Definition: MzIdentMLDOMHandler.h:220
long double fragment_tolerance
Definition: MzIdentMLDOMHandler.h:206
String spectrum_identification_list_ref
Definition: MzIdentMLDOMHandler.h:185
int stop
Definition: MzIdentMLDOMHandler.h:166
std::map< String, size_t > si_pro_map_
mapping SpectrumIdentificationList id -> index to ProteinIdentification in pro_id_
Definition: MzIdentMLDOMHandler.h:240
MzIdentMLDOMHandler & operator=(const MzIdentMLDOMHandler &rhs)
String database_ref
Definition: MzIdentMLDOMHandler.h:175
void parseProteinAmbiguityGroupElement_(xercesc::DOMElement *proteinAmbiguityGroupElement, ProteinIdentification &protein_identification)
void parseSpectrumIdentificationElements_(xercesc::DOMNodeList *spectrumIdentificationElements)
void parsePeptideEvidenceElements_(xercesc::DOMNodeList *peptideEvidenceElements)
CVTerm searchtype
Definition: MzIdentMLDOMHandler.h:199
std::map< String, String > pv_db_map_
mapping PeptideEvidence id -> DBSequence id
Definition: MzIdentMLDOMHandler.h:248
void buildAnalysisDataCollection_(xercesc::DOMElement *analysisElements)
String search_engine_version_
Definition: MzIdentMLDOMHandler.h:229
std::map< String, AnalysisSoftware > as_map_
mapping AnalysisSoftware id -> AnalysisSoftware
Definition: MzIdentMLDOMHandler.h:231
void buildEnclosedCV_(xercesc::DOMElement *parentElement, const String &encel, const String &acc, const String &name, const String &cvref)
bool idec
Definition: MzIdentMLDOMHandler.h:169
ControlledVocabulary unimod_
Controlled vocabulary for modifications (unimod from OpenMS/share/OpenMS/CV/unimod....
Definition: MzIdentMLDOMHandler.h:96
std::map< String, String > xl_mod_map_
mapping peptide id -> cross-linking reagent name
Definition: MzIdentMLDOMHandler.h:261
std::map< String, String > xl_id_donor_map_
mapping Peptide id -> crosslink donor value
Definition: MzIdentMLDOMHandler.h:255
void parseSpectrumIdentificationProtocolElements_(xercesc::DOMNodeList *spectrumIdentificationProtocolElements)
AASequence parsePeptideSiblings_(xercesc::DOMElement *peptide)
String name
Definition: MzIdentMLDOMHandler.h:159
std::multimap< String, String > p_pv_map_
mapping Peptide id -> PeptideEvidence id, multiple PeptideEvidences can have equivalent Peptides.
Definition: MzIdentMLDOMHandler.h:249
void buildInputDataCollection_(xercesc::DOMElement *inputElements)
void writeMzIdentMLFile(const std::string &mzid_file)
Provides the functionality to write a mzid with a handler object.
xercesc::XercesDOMParser mzid_parser_
Definition: MzIdentMLDOMHandler.h:223
CVTermList threshold_cvs
Definition: MzIdentMLDOMHandler.h:207
String fixed_mod
Definition: MzIdentMLDOMHandler.h:190
String accession
Definition: MzIdentMLDOMHandler.h:176
CVTermList specificities
Definition: MzIdentMLDOMHandler.h:194
char pre
Definition: MzIdentMLDOMHandler.h:167
String residues
Definition: MzIdentMLDOMHandler.h:192
std::map< String, DataValue > threshold_ups
Definition: MzIdentMLDOMHandler.h:208
std::pair< CVTermList, std::map< String, DataValue > > parseParamGroup_(xercesc::DOMNodeList *paramGroup)
First: CVparams, Second: userParams (independent of each other)
CVTermList cvs
Definition: MzIdentMLDOMHandler.h:177
std::map< String, AASequence > pep_map_
mapping Peptide id -> Sequence
Definition: MzIdentMLDOMHandler.h:246
void parseAnalysisSoftwareList_(xercesc::DOMNodeList *analysisSoftwareElements)
MzIdentMLDOMHandler(const std::vector< ProteinIdentification > &pro_id, const std::vector< PeptideIdentification > &pep_id, const String &version, const ProgressLogger &logger)
Constructor for a write-only handler for internal identification structures.
MzIdentMLDOMHandler(std::vector< ProteinIdentification > &pro_id, std::vector< PeptideIdentification > &pep_id, const String &version, const ProgressLogger &logger)
Constructor for a read-only handler for internal identification structures.
std::list< std::list< String > > hit_pev_
writing help only
Definition: MzIdentMLDOMHandler.h:252
MzIdentMLDOMHandler(const MzIdentMLDOMHandler &rhs)
std::map< String, SignedSize > xl_donor_pos_map_
mapping donor value -> cross-link modification location
Definition: MzIdentMLDOMHandler.h:258
std::map< String, SpectrumIdentificationProtocol > sp_map_
mapping SpectrumIdentificationProtocol id -> SpectrumIdentificationProtocol
Definition: MzIdentMLDOMHandler.h:243
CVTermList modification_parameter
Definition: MzIdentMLDOMHandler.h:204
void buildCvList_(xercesc::DOMElement *cvElements)
std::map< String, DataValue > parameter_ups
Definition: MzIdentMLDOMHandler.h:202
void buildSequenceCollection_(xercesc::DOMElement *sequenceCollectionElements)
void parseSpectrumIdentificationListElements_(xercesc::DOMNodeList *spectrumIdentificationListElements)
void parseDBSequenceElements_(xercesc::DOMNodeList *dbSequenceElements)
char post
Definition: MzIdentMLDOMHandler.h:168
void buildAnalysisProtocolCollection_(xercesc::DOMElement *protocolElements)
std::map< String, String > xl_id_acceptor_map_
mapping peptide id of acceptor peptide -> crosslink acceptor value
Definition: MzIdentMLDOMHandler.h:257
String spectrum_identification_protocol_ref
Definition: MzIdentMLDOMHandler.h:184
XMLCh * xml_name_attr_ptr_
Definition: MzIdentMLDOMHandler.h:221
virtual ~MzIdentMLDOMHandler()
Destructor.
String spectra_data_ref
Definition: MzIdentMLDOMHandler.h:182
const String schema_version_
Internal version keeping.
Definition: MzIdentMLDOMHandler.h:109
void parsePeptideElements_(xercesc::DOMNodeList *peptideElements)
Struct to hold the used analysis software for that file.
Definition: MzIdentMLDOMHandler.h:158
Struct to hold the information from the DBSequence xml tag.
Definition: MzIdentMLDOMHandler.h:173
Struct to hold the information from the DatabaseInput xml tag.
Definition: MzIdentMLDOMHandler.h:212
Struct to hold the information from the ModificationParam xml tag.
Definition: MzIdentMLDOMHandler.h:189
Struct to hold the PeptideEvidence information.
Definition: MzIdentMLDOMHandler.h:164
Struct to hold the information from the SpectrumIdentification xml tag.
Definition: MzIdentMLDOMHandler.h:181
Struct to hold the information from the SpectrumIdentificationProtocol xml tag.
Definition: MzIdentMLDOMHandler.h:198
Represents the peptide hits for a spectrum.
Definition: PeptideIdentification.h:39
Base class for all classes that want to report their progress.
Definition: ProgressLogger.h:27
Representation of a protein identification run.
Definition: ProteinIdentification.h:50
A more convenient string class.
Definition: String.h:34
Main OpenMS namespace.
Definition: openswathalgo/include/OpenMS/OPENSWATHALGO/DATAACCESS/ISpectrumAccess.h:19
Representation of a CV term.
Definition: ControlledVocabulary.h:35
Search parameters of the DB search.
Definition: ProteinIdentification.h:247