00001 /*
00002 ##############################################################################
00003 # file: ms_ms1quantitation.hpp
00004 # 'msparser' toolkit
00005 # An extracted ion chromatogram
00006 ##############################################################################
00007 # COPYRIGHT NOTICE
00008 # Copyright 1998-2014 Matrix Science Limited  All Rights Reserved.
00009 #
00010 ##############################################################################
00011  * @(#)$Source: parser/inc/ms_ms1quantitation.hpp $
00012  * @(#)$Revision: 211b4f6d8eb604aac8475e054404febcf611b332 | MSPARSER_REL_2_8_1-0-gea32989045 $
00013  * @(#)$Date: 2020-06-04 14:02:19 +0100 $
00014 ##############################################################################
00015  */
00016 
00017 #ifndef MS_MS1QUANTITATION_HPP
00018 #define MS_MS1QUANTITATION_HPP
00019 
00020 
00021 
00022 
00023 // Includes from the standard template library
00024 #include <string>
00025 #include <vector>
00026 
00027 namespace msparser_internal
00028 {
00029     class ms_XMLElement;
00030     class ms_XMLHelper;
00031 }
00032 
00033 namespace matrix_science
00034 {
00035     //=========================================================================
00041     class ms_msquant1_mod_format;
00042     class ms_cache_stream;
00043     class ms_progress_info;
00044     class ms_mascotresults_params;
00045 
00047 
00089     class MS_MASCOTRESFILE_API ms_ms1quantitation :
00090         public ms_customquantitation
00091     {
00092         public:
00094             enum MERGE_QUANT_RESULTS_FLAGS {
00095                 MERGE_QUANT_QUERY_NUMBERS_ARE_FROM_MERGED_RESULTS = 0x0001, 
00096                 MERGE_QUANT_MERGE_COMPONENTS_INTO_QUANT_MATCH     = 0x0002, 
00097                 MERGE_QUANT_USE_COMPONENT_FILE_INDEX_TO_FIND_XIC  = 0x0004, 
00098                 MERGE_QUANT_APPLY_THRESHOLDS_AFTER_MERGE          = 0x0008, 
00099 
00100                 MERGE_QUANT_DEFAULT_PRECURSOR_METHOD              = MERGE_QUANT_APPLY_THRESHOLDS_AFTER_MERGE, 
00101                 MERGE_QUANT_DEFAULT_REPLICATE_METHOD              = (MERGE_QUANT_QUERY_NUMBERS_ARE_FROM_MERGED_RESULTS | MERGE_QUANT_MERGE_COMPONENTS_INTO_QUANT_MATCH | MERGE_QUANT_USE_COMPONENT_FILE_INDEX_TO_FIND_XIC)  
00102             };
00103 
00105 
00114             enum MS1QUANT_BUGFIX_NUM {
00115                 BUGFIX_12409 = 12409
00116             };
00117 
00119             ms_ms1quantitation(const ms_peptidesummary & summary, const ms_quant_method & method);
00120 
00121             ~ms_ms1quantitation();
00122 
00124             int getNumPeptideStatuses() const;
00126             ms_ms1quant_peptide_status getPeptideStatus(int index) const;
00127 
00129             int getNumHits() const;
00131             ms_protein * getHit(int index1) const;
00132 
00134             int getNumProteinFamilies() const;
00135 
00137             int getNumMatches() const;
00139             ms_ms1quant_match getMatch(int index) const;
00141             ms_ms1quant_match getMatch(const ms_peptide_quant_key & key) const;
00142 
00143             void clearMatches();            
00144 
00146             void addMatchStatus(
00147                     matrix_science::ms_peptide * peptide,
00148                     matrix_science::ms_ms1quant_peptide_status::peptideStatus status,
00149                     matrix_science::ms_protein * hit = 0);
00150 
00152             void addMatchPeptide(
00153                     matrix_science::ms_protein * hit,
00154                     matrix_science::ms_peptide * peptide,
00155                     const matrix_science::ms_quant_component & component,
00156                     const std::string & matchBelongsToComponent);
00157 
00159             matrix_science::ms_ms1quant_match addMatchInferred(
00160                     const ms_ms1quant_match & siblingMatch,
00161                     int newCharge);
00162 
00164             void removeMatch(matrix_science::ms_peptide_quant_key matchKey);            
00165 
00167             bool findPeptideStatus(int query, int rank, ms_ms1quant_peptide_status * peptideStatus) const;
00168 
00170             bool findMatch(matrix_science::ms_peptide_quant_key matchKey, ms_ms1quant_match * match) const;
00171 
00173             const ms_ms1quant_average_results * getAverageResults() const;
00174 
00176             bool loadXmlFile(const std::string & filename, const std::string & xmlSchemaPath);
00177 
00179             bool saveXmlFile(const std::string & xmlFilename, bool completeReport);
00180 
00182             bool loadCdbFile(const std::string & cdbFilename, const std::string & cacheFilename, bool revalidateQualityThresholds=true);
00183 
00185             bool saveCdbFile(const std::string & cdbFilename, const std::string & cacheFilename, bool createEmptyMissingComponents = false);
00186 
00188             const ms_peptidesummary & getSummary() const;
00189 
00191             std::string getReadableVarMods(std::string varmods);
00192 
00194             std::string getReadableVarModsConcise(std::string varmods);
00195 
00197             std::string getLabelFree(std::string peptideStr, std::string varMods, std::string summedMods, const matrix_science::ms_quant_component & comp) const;
00198 
00200             int getNumMatchesForHit(int hitId, int memberNumber) const;
00202             ms_ms1quant_match getMatchForHit(int index, int hitId, int memberNumber) const;
00203 
00205             int getNumGroups() const;
00206 
00207             int getDatasourceForGroup(int groupIndex1) const;
00208 
00210             int getNumMatchesForGroup(int groupIndex1) const;
00211 
00212             ms_ms1quant_match getMatchForGroup(int matchIndex1, int groupIndex1) const;
00213 
00215             virtual ms_protein_quant_ratio getProteinRatio(const std::string &accession, int dbIdx, const std::string &ratioName) const;
00216 
00218             virtual ms_protein_quant_ratio getProteinRatio(const std::string &accession, int dbIdx, const std::string &ratioName, const ms_peptide_quant_key_vector &includeKeys, const ms_peptide_quant_key_vector &excludeKeys) const;
00219 
00221             std::vector<double> getAllProteinRatios(const std::string & accession, const int dbIdx, const std::string & ratioName) const;
00222 
00224             std::vector<std::vector<bool> > getAllProteinRatioFlags(const std::string & accession, const int dbIdx, const std::string & ratioName) const;
00225 
00227             std::vector<std::vector<double> > getAllProteinRatioValues(const std::string & accession, const int dbIdx, const std::string & ratioName) const;
00228 
00230             std::vector<std::string> getAllProteinRatioPepSeqs(const std::string & accession, const int dbIdx, const std::string & ratioName) const;
00231 
00233             bool isAverageProtocol() const;
00235             void setAverageReferenceProtein(const char *accession, int dbIdx, bool isReferenceUserSelected);
00237             void setAverageNumPeptides(unsigned int num = 3);
00239             void setAverageSelection(const char * value);
00240 #ifndef SWIG
00241 
00242             bool getAveragePeptideIntensity(
00243                     const std::string &accession,
00244                     int dbIdx,
00245                     double * averageIntensity, // average of grouped intensities (each peptide ratio group summed and then average of those sums)
00246                     int * sampleSize, // number of different peptide groups available (after grouping by selection type)
00247                     ms_peptide_quant_key_vector * includeKeys, // peptide ratios selected (may be greater then sampleSize)
00248                     ms_peptide_quant_key_vector * excludeKeys // peptide ratios not selected
00249                     ) const;
00250 #else // SWIG Multiple return values
00251             bool getAveragePeptideIntensity(
00252                 const std::string & accession,
00253                 int                 dbIdx,
00254                 double *            OUTPUT,
00255                 int *               OUTPUT, 
00256                 ms_peptide_quant_key_vector * includeKeys,
00257                 ms_peptide_quant_key_vector * excludeKeys
00258                 ) const;
00259 #endif
00260 
00261             std::vector<int> getQuantitationDatasources() const;
00262 
00263             int getNumMatchedPeptideStatuses() const;
00264 
00266             void initGroupChargeStates(
00267                     std::set<int> * calc_chargestates,
00268                     int iGroup1,
00269                     int minCharge,
00270                     int maxCharge);            
00271 
00272             void calc_complete();
00273 
00275             bool isDataCached(MS1QUANT_BUGFIX_NUM bugNum) const;
00276 
00278             using ms_quantitation::countDistinctPeptides;
00279 
00280 #ifndef SWIG
00281 
00282             virtual int countDistinctPeptides(const std::vector<ms_peptide_quant_key> & activeKeys) const;
00283 #endif
00284 
00285             static matrix_science::ms_ms1quantitation * openQuantitationCdbResults(
00286                 const matrix_science::ms_peptidesummary & summary, 
00287                 const matrix_science::ms_quant_method & method, 
00288                 const std::string & cdbFilename, 
00289                 const std::string & cacheFilename, 
00290                 const matrix_science::ms_quant_method * originalMethod = 0);
00291 
00293             static matrix_science::ms_ms1quantitation * openQuantitationXmlResults(
00294                 const matrix_science::ms_peptidesummary & summary, 
00295                 const matrix_science::ms_quant_method & method,
00296                 const std::string & filename,
00297                 const std::string & xmlSchemaPath="",
00298                 const matrix_science::ms_quant_method * originalMethod = 0);
00299 
00301             bool mergeQuantitationResults(const std::vector<std::string> & cdbFilenames, 
00302                                           const std::vector<std::string> & cacheFilenames,
00303                                           const MERGE_QUANT_RESULTS_FLAGS flags,
00304                                           const matrix_science::ms_quant_method * originalMethod = 0);
00305 
00307             bool mergeQuantitationResults(const std::vector<std::string> & xmlFilenames,
00308                                           const MERGE_QUANT_RESULTS_FLAGS flags,
00309                                           const std::string & xmlSchemaPath="",
00310                                           const matrix_science::ms_quant_method * originalMethod = 0);
00311 #ifndef SWIG
00312 
00313             static bool readResultCounters(const std::string & cdbFile, 
00314                 int &numMatches, int &numHits, int &numPepGrp, int &numPeptideComponents);            
00315 #else // SWIG Multiple return values
00316             static bool readResultCounters(const std::string & cdbFile,
00317                 int &OUTPUT, int &OUTPUT, int &OUTPUT, int &OUTPUT);
00318 #endif
00319 
00320         protected:
00321 
00322         private:
00323             //Do not copy this object
00324             ms_ms1quantitation(const ms_ms1quantitation & src);
00325             ms_ms1quantitation& operator=(const ms_ms1quantitation & right);
00326 
00327             void initialiseAfterLoad();    
00328             // normalisation
00329             std::list<ms_ms1quant_match> selectMatchesForNormalisation(const ms_quant_normalisation *norm);
00330             double calculatePeptideNormalisationBase(const std::string &ratioName) const;
00332             void normalisePeptideRatios();
00333             // end normalisation
00334 
00335             bool saveHitToCdb(matrix_science::ms_tinycdb & cdbFile, const matrix_science::ms_protein * protein, int streamNumber) const;
00336             bool saveHitToStream(matrix_science::ms_cache_stream & stream, const matrix_science::ms_protein * protein) const;
00337             bool saveHitToXml(msparser_internal::ms_XMLHelper & helper, msparser_internal::ms_XMLElement & element, const matrix_science::ms_protein * protein, int hitId, int memberNumber);            
00338 
00339             void getProteinRatioOverrideKeys(
00340                     ms_peptide_quant_key_vector * incFK,
00341                     ms_peptide_quant_key_vector * excFK,
00342                     const std::string & ratioName,
00343                     int hitId,
00344                     int memberNumber) const;
00345 
00346             ms_protein_quant_ratio getProteinAverageIntensityRatio(
00347                     const std::string &accession,
00348                     int dbIdx
00349                     ) const;
00350 
00351             void rematchInferredMatches(const bool allowMassTimeMatches);  // populates inferredMatchSource_
00352             void updatePeptideStatus(matrix_science::ms_ms1quant_peptide_status status, const matrix_science::ms_quant_quality * quality);
00353             void updateMatches(const matrix_science::ms_quant_method * method);           // removes any ms1 quant matches which don't have any valid ms2 peptide matches
00354             void updateMatchPeptideEvidence();  // updates the ms2 matches associated (or not associated) with an ms1 quant match
00355             bool updateProteinMatches();        // ensures only valid anchor proteins are returned
00356 
00357             // Can the new settings be handled my ms_ms1quantitation, or do additional values need to be calculated by MsQuantLib
00358             static bool canHandleSettings(const matrix_science::ms_quant_method * originalMethod, const matrix_science::ms_quant_method * newMethod, matrix_science::ms_ms1quantitation * quantResults);
00359 
00360             std::map<matrix_science::ms_peptide_quant_key, ms_ms1quant_peptide_status_body> getMatchedPeptideStatuses() const;            
00361 
00363             std::vector<ms_ms1quant_match> getMatchesForHit(int hitId, int memberNumber) const;
00364 
00366             bool saveGroupToXml(msparser_internal::ms_XMLHelper & helper, msparser_internal::ms_XMLElement & element, const matrix_science::ms_protein * protein, int hitId, int memberNumber) const;
00367             void reorderMatchesAfterMerge();
00368 
00370             bool mergeCdbFiles(const std::vector<std::string> & cdbFilenames,
00371                 const std::vector<std::string> & cacheFilenames,
00372                 const std::vector<std::string> & resultFilenames,
00373                 const matrix_science::ms_mascotresults_params & parameters,
00374                 const std::string & cacheDirectory,
00375                 const MERGE_QUANT_RESULTS_FLAGS flags);
00376 
00378             bool mergeXmlFiles(
00379                 const std::vector<std::string> & xmlFilenames,
00380                 const std::vector<std::string> & resultFilenames,
00381                 const std::string & cacheDirectory,
00382                 const std::string & xmlSchemaPath,
00383                 const MERGE_QUANT_RESULTS_FLAGS flags);
00384 
00386             void getKeyStrings(const int dataSourceId, const std::string & peptideStr, const std::string & labelFreeVarMods, const int chargeState,
00387                                std::string & groupKey, std::string & uniqueKey) const;
00388 
00389             const ms_peptidesummary & peptideSummary_;
00390             int nextMatchId_;
00391             std::map<ms_peptide_quant_key, ms_ms1quant_match> matches_; // matchId -> match
00392             std::map<std::pair<int, int>, ms_ms1quant_peptide_status> peptideStatuses_; // [query, rank] -> status
00393             std::map<std::string, ms_peptide_quant_key> matchKeys_; // peptideStr:datasourceId:labelFree:charge -> matchId
00394             std::map<std::string, std::set<ms_peptide_quant_key> > matchGroups_; // "datasource:peptideStr:labelFree" -> matchId[]
00395             std::vector<std::string> matchGroupNames_; // names in matchGroups in order of addition (to match with processing order in Distiller)
00396 
00397             
00398             // matchId -> matchId of the 'source/sibling' identified match - always the most intense sibling if there is more than one possibility
00399             std::map<ms_peptide_quant_key, ms_peptide_quant_key> inferredMatchSource_;  
00400 
00401             typedef std::pair<int, int> Hitkey; // hitId, memberNumber (0 if not a family)
00402             std::vector<Hitkey> hitkeys_;
00403 
00404             bool hasAverageResults_;
00405             ms_ms1quant_average_results averageResults_;
00406 
00407             matrix_science::ms_cache_file cacheFile_;
00408 
00409             mutable const ms_msquant1_mod_format * modFormat_;            
00410 
00411             std::vector<matrix_science::ms_cache_file *> multifileCacheFiles_;
00412 
00413             std::string bugsfixed_;
00414     };
00415  // end of quantitation_group
00417     //=========================================================================
00418 
00419 }   // matrix_science namespace
00420 
00421 
00422 #endif // MS_MS1QUANTITATION_HPP
00423 
00424 /*------------------------------- End of File -------------------------------*/