WIP : writing MassChroqML

src/core/peptidextp.cpp

@@ -67,8 +67,12 @@ PeptideXtpSp PeptideXtp::makePeptideXtpSp() const {
 
 
 pappso::mz PeptideXtp::getGroupingMass() const {
+    return getNativePeptideP()->getMass();
+}
+
+const pappso::Peptide * PeptideXtp::getNativePeptideP() const {
     if (_sp_native_peptide.get() == nullptr) {
-        return getMass();
+        return this;
     }
-    return _sp_native_peptide.get()->getMass();
+    return _sp_native_peptide.get();
 }

src/core/peptidextp.h

@@ -40,12 +40,15 @@ public:
     ~PeptideXtp();
     
     PeptideXtpSp makePeptideXtpSp() const;
+    
     /** \brief get the tehoretical mass to use for grouping
      * This mass might be different than the true peptide mass to recognize that a tagged 
      * peptide with taget modification is a light, inter or heavy version of the same peptide
  */
     pappso::mz getGroupingMass() const;
     
+    const pappso::Peptide * getNativePeptideP() const;
+    
     /** \brief human readable string that contains modifications
      * */
     const QString getModifString() const;

src/output/masschroqml.cpp

@@ -32,6 +32,15 @@
 #include <pappsomspp/utils.h>
 #include <pappsomspp/grouping/grpprotein.h>
 
+
+struct McqPeptide {
+    QString id;
+    pappso::pappso_double mh;
+    QStringList mods;
+    QStringList prot_ids;
+    QString seq;
+};
+
 MassChroQml::MassChroQml(const QString & out_filename)
 {
     //_p_digestion_pipeline = p_digestion_pipeline;
@@ -116,29 +125,58 @@ void MassChroQml::write(ProjectSp sp_project) {
 void MassChroQml::writePeptideList() {
 //<peptide_list>
     _output_stream->writeStartElement("peptide_list");
-    QStringList empty;
-    for (const GroupingGroupSp & group:_group_list) {
-        std::map<const PeptideMatch *, QStringList> peptide_sg_map;
-        unsigned int group_number = group.get()->getGroupNumber();
-        for (auto && pair_peptide_match :group.get()->getPairSgNumberPeptideMatchList()) {
-            unsigned int sg_number = pair_peptide_match.first;
-            const PeptideMatch * p_peptide_match = pair_peptide_match.second;
-            if (!p_peptide_match->isGrouped()) continue;
-            
-            std::pair<std::map< const PeptideMatch *, QStringList>::iterator,bool> ret = peptide_sg_map.insert(std::pair<const PeptideMatch *, QStringList>(p_peptide_match, empty));
-
-            QString sg_str = QString("%1.%2.a1").arg(pappso::Utils::getLexicalOrderedString(group_number)).arg(pappso::Utils::getLexicalOrderedString(sg_number));
-            if (!ret.first->second.contains(sg_str)) {
-                ret.first->second << sg_str;
+    std::vector<McqPeptide> mcq_peptide_list;
+    for (ProteinMatch * p_protein_match :_p_identification_group->getProteinMatchList()) {
+        if (!p_protein_match->isGrouped()) continue;
+        for (PeptideMatch * peptide_match :p_protein_match->getPeptideMatchList()) {
+            if (!peptide_match->isGrouped()) continue;
+
+            McqPeptide mcq_peptide;
+            mcq_peptide.id = peptide_match->getGrpPeptideSp().get()->getGroupingId();
+            mcq_peptide.mods << peptide_match->getPeptideXtpSp().get()->getModifString();
+            mcq_peptide.prot_ids << p_protein_match->getGrpProteinSp().get()->getGroupingId();
+            mcq_peptide.seq = peptide_match->getPeptideXtpSp().get()->getSequence();
+            mcq_peptide.mh = peptide_match->getPeptideXtpSp().get()->getNativePeptideP()->getMz(1);
+            mcq_peptide_list.push_back(mcq_peptide);
+        }
+
+    }
+
+    //sort list
+    std::sort(mcq_peptide_list.begin(),mcq_peptide_list.end(),[](const McqPeptide & first, const McqPeptide & second) {
+        return (first.id < second.id) ;
+    });
+
+    std::vector<McqPeptide> cumul_mcq_peptide_list;
+    if (mcq_peptide_list.size() > 0) {
+        McqPeptide cumul = mcq_peptide_list[0];
+        for (McqPeptide & mcq_peptide : mcq_peptide_list) {
+            if (cumul.id == mcq_peptide.id) {
+                if (!cumul.prot_ids.contains(mcq_peptide.prot_ids[0])) {
+                    cumul.prot_ids << mcq_peptide.prot_ids[0];
+                }
+                if (!cumul.mods.contains(mcq_peptide.mods[0])) {
+                    cumul.mods << mcq_peptide.mods[0];
+                }
+            }
+            else {
+                cumul_mcq_peptide_list.push_back(cumul);
+                cumul = mcq_peptide;
             }
         }
+        cumul_mcq_peptide_list.push_back(cumul);
+    }
+
 
-        for (auto && pair_peptide_sgid : peptide_sg_map) {
+    for (McqPeptide mcq_peptide : cumul_mcq_peptide_list) {
         //<peptide id="pep0" mh="1463.626" mods="114.08" prot_ids="P1.1"
         //	seq="TCVADESHAGCEK">
         _output_stream->writeStartElement("peptide");
-        _output_stream->writeAttribute("id",pair_peptide_sgid.first->getGrpPeptideSp().get()->getGroupingId());
-        _output_stream->writeAttribute("prot_ids",pair_peptide_sgid.second.join(" "));
+        _output_stream->writeAttribute("id",mcq_peptide.id);
+        _output_stream->writeAttribute("mods",mcq_peptide.mods.join("|"));
+        _output_stream->writeAttribute("prot_ids",mcq_peptide.prot_ids.join(" "));
+        _output_stream->writeAttribute("mh",QString::number(mcq_peptide.mh, 'f', 10));
+        _output_stream->writeAttribute("seq",mcq_peptide.seq);
         /*
         <modifications><!-- this tag is optional but gives an exact mass computation -->
             <psimod at="2" acc="MOD:00397"></psimod>
@@ -148,7 +186,6 @@ void MassChroQml::writePeptideList() {
         <observed_in data="samp1" scan="798" z="2" />*/
         //</peptide>
         _output_stream->writeEndElement();
-        }
     }
     //</peptide_list>
     _output_stream->writeEndElement();
@@ -187,7 +224,6 @@ void MassChroQml::writeProteinList() {
         if (!p_protein_match->isGrouped()) continue;
         //<protein desc="conta|P02769|ALBU_BOVIN SERUM ALBUMIN PRECURSOR."
         //	id="P1.1" />
-        _group_list.insert(p_protein_match->getGroupingGroupSp());
         pappso::GrpProteinSp grp_protein = p_protein_match->getGrpProteinSp();
         if (grp_protein.get()->getRank() == 1) {
             _output_stream->writeStartElement("protein");

src/output/masschroqml.h

@@ -53,7 +53,6 @@ private :
     QXmlStreamWriter * _output_stream;
     ProjectSp _sp_project;
     IdentificationGroup * _p_identification_group;
-    std::set<GroupingGroupSp> _group_list;
 };
 
 #endif // MASSCHROQML_H