WIP: using pwiz to read identification files

src/core/identification_sources/identificationpwizfile.cpp

@@ -36,6 +36,7 @@
 IdentificationPwizFile::IdentificationPwizFile(const QFileInfo & ident_file) : IdentificationDataSource(ident_file.absoluteFilePath()), _ident_file(ident_file)
 {
     //_engine = IdentificationEngine::XTandem;
+  _engine = IdentificationEngine::peptider;
 }
 
 IdentificationPwizFile::IdentificationPwizFile(const IdentificationPwizFile& other) : IdentificationDataSource(other),_ident_file (other._ident_file)
@@ -60,7 +61,7 @@ pappso::SpectrumSp IdentificationPwizFile::getSpectrumSp(unsigned int scan_numbe
 
 
 void IdentificationPwizFile::parseTo(Project* p_project) {
-    qDebug() << "Project::readXpipFile begin";
+    qDebug() << "IdentificationPwizFile::parseTo begin";
 
     IdentificationPwizReader pwiz_reader(_ident_file);
 
@@ -91,6 +92,7 @@ void IdentificationPwizFile::parseTo(Project* p_project) {
     pwiz_reader.read(this, p_project, identification_group_p);
 
 
+    qDebug() << "IdentificationPwizFile::parseTo end";
       //  throw pappso::PappsoException(QObject::tr("Error reading %1 XPIP file :\n %2").arg(_xtandem_file.absoluteFilePath()).arg(parser->errorString()));
     
 

src/core/identification_sources/identificationxtandemfile.cpp

@@ -52,7 +52,7 @@ pappso::SpectrumSp IdentificationXtandemFile::getSpectrumSp(unsigned int scan_nu
 
 
 void IdentificationXtandemFile::parseTo(Project* p_project) {
-    qDebug() << "Project::readXpipFile begin";
+    qDebug() << "IdentificationXtandemFile::parseTo begin";
 
     qDebug() << "Read X!Tandem XML result file '" << _xtandem_file.absoluteFilePath() << "'";
 
@@ -104,4 +104,5 @@ void IdentificationXtandemFile::parseTo(Project* p_project) {
         throw pappso::PappsoException(QObject::tr("Error reading %1 XPIP file :\n %2").arg(_xtandem_file.absoluteFilePath()).arg(parser->errorString()));
     }
 
+    qDebug() << "IdentificationXtandemFile::parseTo end";
 }

src/core/peptidematch.cpp

@@ -32,6 +32,21 @@ PeptideMatch::PeptideMatch(MsRun * msrunid_sp, unsigned int scan) {
     _hash_sample_scan = PeptideMatch::_hash_fn (QString("%1 %2").arg(msrunid_sp->getXmlId()).arg(_scan).toStdString());
 }
 
+PeptideMatch::PeptideMatch(const PeptideMatch & other): PeptideMatch(other._msrunid_sp, other._scan) {
+    _hash_sample_scan = other._hash_sample_scan;
+    _sp_grp_peptide = other._sp_grp_peptide;
+    _peptide_sp = other._peptide_sp;
+    _rt = other._rt;
+    _evalue = other._evalue;
+    _exp_mass = other._exp_mass;
+    _start=other._start;
+    _charge= other._charge;
+    _p_identification_source = other._p_identification_source;
+    _params = other._params;
+    _checked = other._checked;
+    _proxy_valid = other._proxy_valid;
+}
+
 void PeptideMatch::updateAutomaticFilters(const AutomaticFilterParameters & automatic_filter_parameters) {
     _proxy_valid = false;
 

src/core/peptidematch.h

@@ -35,6 +35,7 @@ class PeptideMatch
 {
 public :
     PeptideMatch(MsRun * msrunid_sp, unsigned int scan);
+    PeptideMatch(const PeptideMatch & other);
     std::size_t getHashSampleScan() const;
     std::size_t getHashPeptideMassSample() const;
 

src/core/project.cpp

@@ -24,6 +24,7 @@
 #include "../input/xpipsaxhandler.h"
 #include "peptidematch.h"
 #include "proteinmatch.h"
+#include <QDebug>
 
 Project::Project()
 {
@@ -46,6 +47,7 @@ void Project::readResultFile(QString filename) {
     IdentificationDataSourceSp ident_source = _identification_data_source_store.getInstance(filename);
 
     ident_source.get()->parseTo(this);
+    qDebug() << "Project::readResultFile end";
 }
 
 bool Project::isCombineMode() const {
@@ -92,11 +94,14 @@ const AutomaticFilterParameters & Project::getAutomaticFilterParameters() const
     return _automatic_filter_parameters;
 }
 void Project::updateAutomaticFilters(const AutomaticFilterParameters & automatic_filter_parameters) {
+  
+    qDebug() << "Project::updateAutomaticFilters begin";
     _automatic_filter_parameters = automatic_filter_parameters;
     for (auto & p_id_group : _identification_goup_list) {
         p_id_group->updateAutomaticFilters(_automatic_filter_parameters);
     }
 
+    qDebug() << "Project::updateAutomaticFilters end";
 }
 ProjectSp Project::makeProjectSp() const {
     return std::make_shared<Project>(*this);

src/gui/workerthread.cpp

@@ -41,7 +41,7 @@
 WorkerThread::WorkerThread(MainWindow * p_main_window)
 {
     qDebug() << "WorkerThread::WorkerThread begin MainWindow";
-    
+
     _p_work_monitor = new WorkMonitor();
 
 #if QT_VERSION >= 0x050000
@@ -53,7 +53,7 @@ WorkerThread::WorkerThread(MainWindow * p_main_window)
     */
 #else
 // Qt4 code
-    //worker message 
+    //worker message
     connect(_p_work_monitor, SIGNAL(workerMessage(QString)), p_main_window,SLOT(doDisplayLoadingMessage(QString)));
     connect(_p_work_monitor, SIGNAL(workerMessage(QString, int)), p_main_window,SLOT(doDisplayLoadingMessage(QString, int)));
 
@@ -87,7 +87,7 @@ WorkerThread::WorkerThread(MainWindow * p_main_window)
 WorkerThread::WorkerThread(ProjectWindow * p_project_window)
 {
     qDebug() << "WorkerThread::WorkerThread begin ProjectWindow";
-       
+
     _p_work_monitor = new WorkMonitor();
 
 #if QT_VERSION >= 0x050000
@@ -95,8 +95,8 @@ WorkerThread::WorkerThread(ProjectWindow * p_project_window)
 #else
 // Qt4 code
     qDebug() << "WorkerThread::WorkerThread Qt4 code";
-    
-        //worker message 
+
+    //worker message
     connect(_p_work_monitor, SIGNAL(workerMessage(QString)), p_project_window,SLOT(doDisplayLoadingMessage(QString)));
     connect(_p_work_monitor, SIGNAL(workerMessage(QString, int)), p_project_window,SLOT(doDisplayLoadingMessage(QString, int)));
 
@@ -104,16 +104,16 @@ WorkerThread::WorkerThread(ProjectWindow * p_project_window)
     //grouping
     connect(p_project_window, SIGNAL(operateGrouping(ProjectSp)), this,SLOT(doGrouping(ProjectSp)));
     connect(this, SIGNAL(groupingFinished()), p_project_window,SLOT(doGroupingFinished()));
- 
-    
-   //PTM grouping on IdentificationGroup
+
+
+    //PTM grouping on IdentificationGroup
     connect(p_project_window, SIGNAL(operatePtmGroupingOnIdentification(IdentificationGroup *)), this,SLOT(doPtmGroupingOnIdentification(IdentificationGroup *)));
     connect(this, SIGNAL(ptmGroupingOnIdentificationFinished(IdentificationGroup *)), p_project_window,SLOT(refreshPtmGroup(IdentificationGroup *)));
 
     //grouping on IdentificationGroup
     connect(p_project_window, SIGNAL(operateGroupingOnIdentification(IdentificationGroup *, GroupingType)), this,SLOT(doGroupingOnIdentification(IdentificationGroup *, GroupingType)));
     connect(this, SIGNAL(groupingOnIdentificationFinished(IdentificationGroup *)), p_project_window,SLOT(refreshGroup(IdentificationGroup *)));
-    
+
     connect(this, SIGNAL(operationFailed(QString)), p_project_window,SLOT(doOperationFailed(QString)));
     connect(this, SIGNAL(operationFinished()), p_project_window,SLOT(doOperationFinished()));
 
@@ -277,6 +277,7 @@ void WorkerThread::doLoadingResults(bool is_individual, AutomaticFilterParameter
 
         emit loadingMessage(tr("filtering proteins"));
         project_sp.get()->updateAutomaticFilters(param);
+        qDebug() << "WorkerThread::doLoadingResults emit projectReady(project_sp) ";
         emit projectReady(project_sp);
 
     }

src/input/identificationpwizreader.cpp

@@ -35,6 +35,7 @@
 #include <QDir>
 #include <QString>
 #include <pappsomspp/pappsoexception.h>
+#include <pappsomspp/exception/exceptionoutofrange.h>
 
 #include <pwiz/data/identdata/IdentData.hpp>
 #include <pwiz/data/identdata/IdentDataFile.hpp>
@@ -98,7 +99,7 @@ void IdentificationPwizReader::read(
     try {
         ProteinStore & protein_store = _p_project->getProteinStore();
         PeptideStore & peptide_store = _p_project->getPeptideStore();
-        std::map<QString, ProteinXtpSp> map_id2protein;
+        std::map<QString, ProteinMatch *> map_id2protein;
         std::map<QString, PeptideXtpSp> map_id2peptide;
 
         for (pwiz::identdata::DBSequencePtr seq_ptr :_pwiz_ident_data_file->sequenceCollection.dbSequences) {
@@ -109,7 +110,15 @@ void IdentificationPwizReader::read(
             protein.setSequence(QString::fromStdString( sequence_pwiz->seq));
             ProteinXtpSp protein_sp(protein.makeProteinXtpSp());
             protein_sp = protein_store.getInstance(protein_sp);
-            map_id2protein.insert(std::pair<QString, ProteinXtpSp>(QString::fromStdString( sequence_pwiz->id), protein_sp));
+
+
+            ProteinMatch * protein_match_p = _p_identification_group->getProteinMatchInstance(protein_sp.get()->getAccession());
+            protein_match_p->setProteinXtpSp(protein_sp);
+	    protein_match_p->setChecked(true);
+
+
+            map_id2protein.insert(std::pair<QString, ProteinMatch *>(QString::fromStdString( sequence_pwiz->id), protein_match_p));
+
         }
 
         for (pwiz::identdata::PeptidePtr seq_ptr :_pwiz_ident_data_file->sequenceCollection.peptides) {
@@ -136,7 +145,6 @@ void IdentificationPwizReader::read(
                     throw pappso::PappsoException(QObject::tr("Scan number not found"));
                 }
                 //<cvParam accession="MS:1001115" cvRef="PSI-MS" value="4925" name="scan number(s)"/>
-               PeptideMatch * p_peptide_match = new PeptideMatch(_sp_msrun.get(), scan);
 
                 double  rt=0;
 
@@ -144,30 +152,61 @@ void IdentificationPwizReader::read(
                 {
                     rt =
                         std::stod(spectrum_ptr.get()->cvParam(pwiz::cv::MS_retention_time).value);
-                        p_peptide_match->setRetentionTime(rt);
                 }
                 else {
                     //throw pappso::PappsoException(QObject::tr("Scan number not found"));
                 }
-                /*
-                p_peptide_match->setEvalue(attributes.value("expect").simplified().toDouble());
-                //qDebug() << "XtandemSaxHandler::startElement_domain evalue "  << _p_peptide_match->getEvalue() << " scan " << _p_peptide_match->getScan();
 
-                pappso::pappso_double xtandem_mhtheo = attributes.value("mh").simplified().toDouble();
-                pappso::pappso_double xtandem_delta = attributes.value("delta").simplified().toDouble();
-
-                pappso::pappso_double exp_mass = xtandem_mhtheo + xtandem_delta - pappso::MHPLUS;
-
-                p_peptide_match->setExperimentalMass(exp_mass);
-                p_peptide_match->setStart(attributes.value("start").simplified().toUInt()-1);
-                p_peptide_match->setCharge(_charge);
-
-                p_peptide_match->setParam(PeptideMatchParam::tandem_hyperscore,  QVariant( attributes.value("hyperscore").toDouble()));
-
-                p_peptide_match->setIdentificationDataSource( _p_identification_data_source);
-                p_peptide_match->setChecked(true);
-                _p_protein_match->addPeptideMatch(p_peptide_match);
-                */
+                //<SpectrumIdentificationItem passThreshold="true" rank="1"
+                //experimentalMassToCharge="388.2189331" id="sii2" chargeState="2" calculatedMassToCharge="388.2161235">
+
+                //      <PeptideEvidenceRef peptideEvidence_ref="dbseq821_pep947_414"/>
+                //    <PeptideEvidenceRef peptideEvidence_ref="dbseq821_pep947_414"/>
+                //  <cvParam accession="MS:1001330" cvRef="PSI-MS" value="0.000452757388" name="X!Tandem:expect"/>
+                //<cvParam accession="MS:1001331" cvRef="PSI-MS" value="24.275219" name="X!Tandem:hyperscore"/>
+                //    </SpectrumIdentificationItem>
+                for ( pwiz::identdata::SpectrumIdentificationItemPtr spid_item : spectrum_ptr.get()->spectrumIdentificationItem) {
+                    PeptideMatch * p_peptide_match = new PeptideMatch(_sp_msrun.get(), scan);
+                    p_peptide_match->setRetentionTime(rt);
+                    p_peptide_match->setCharge(spid_item->chargeState);
+                    pappso::pappso_double exp_mass = spid_item->experimentalMassToCharge * ((double) spid_item->chargeState) - (((double) spid_item->chargeState) * pappso::MHPLUS);
+                    p_peptide_match->setExperimentalMass(exp_mass);
+                    p_peptide_match->setIdentificationDataSource( p_identification_data_source);
+
+                    p_peptide_match->setChecked(true);
+                    if (spectrum_ptr.get()->hasCVParam(pwiz::cv::MS_X_Tandem_expect))
+                    {
+                        p_peptide_match->setEvalue(std::stod(spectrum_ptr.get()->cvParam(pwiz::cv::MS_X_Tandem_expect).value));
+                    }
+                    if (spectrum_ptr.get()->hasCVParam(pwiz::cv::MS_X_Tandem_hyperscore))
+                    {
+                        p_peptide_match->setParam(PeptideMatchParam::tandem_hyperscore,  QVariant(std::stod(spectrum_ptr.get()->cvParam(pwiz::cv::MS_X_Tandem_hyperscore).value)));
+                    }
+
+                    for ( pwiz::identdata::PeptideEvidencePtr peptide_evidence_sp : spid_item.get()->peptideEvidencePtr) {
+                        PeptideMatch * p_peptide_match_from_evidence = new PeptideMatch(*p_peptide_match);
+                        p_peptide_match_from_evidence->setStart(peptide_evidence_sp.get()->start);
+
+
+                        auto it_pep = map_id2peptide.find(QString::fromStdString(peptide_evidence_sp.get()->peptidePtr.get()->id));
+                        if (it_pep == map_id2peptide.end()) {
+                            throw pappso::ExceptionOutOfRange(QObject::tr("key %1 not found in map_id2peptide, reading SpectrumIdentificationItem %2")
+                                                              .arg(QString::fromStdString(peptide_evidence_sp.get()->peptidePtr.get()->id))
+                                                              .arg(QString::fromStdString(spid_item.get()->id))
+                                                             );
+                        }
+                        p_peptide_match->setPeptideXtpSp(it_pep->second);
+
+                        auto it = map_id2protein.find(QString::fromStdString(peptide_evidence_sp.get()->dbSequencePtr.get()->id));
+                        if (it == map_id2protein.end()) {
+                            throw pappso::ExceptionOutOfRange(QObject::tr("key %1 not found in map_id2protein").arg(QString::fromStdString(peptide_evidence_sp.get()->dbSequencePtr.get()->id)));
+                        }
+                        ProteinMatch * protein_match_p = it->second;
+
+                        protein_match_p->addPeptideMatch(p_peptide_match_from_evidence);
+                    }
+                    delete p_peptide_match;
+                }
             }
         }