coding variable modifications on peptides (MASCOT)

src/core/proteinxtp.cpp

@@ -187,7 +187,7 @@ void ProteinXtp::parseAccession2dbxref() {
 
     for (unsigned int i=0; i < access_list.size(); i++) {
         accession = access_list.at(i);
-        qDebug() << "ProteinXtp::parseAccession2dbxref accession " << accession;
+        //qDebug() << "ProteinXtp::parseAccession2dbxref accession " << accession;
         if (atg.indexIn(accession, 0) != -1) {
             QStringList temp = accession.split(".");
             _dbxref_list.push_back(DbXref(ExternalDatabase::AGI_LocusCode, temp.at(0)));
@@ -195,13 +195,13 @@ void ProteinXtp::parseAccession2dbxref() {
         }
 
         if (swiss_prot.indexIn(accession, 0) != -1) {
-            qDebug() << "ProteinXtp::parseAccession2dbxref accession SwissProt " << accession;
+            //qDebug() << "ProteinXtp::parseAccession2dbxref accession SwissProt " << accession;
             _dbxref_list.push_back(DbXref(ExternalDatabase::SwissProt, accession));
             continue;
         }
 
         if (trembl.indexIn(accession, 0) != -1) {
-            qDebug() << "ProteinXtp::parseAccession2dbxref accession TrEMBL " << accession;
+            //qDebug() << "ProteinXtp::parseAccession2dbxref accession TrEMBL " << accession;
             _dbxref_list.push_back(DbXref(ExternalDatabase::TrEMBL, accession));
             continue;
         }
@@ -236,7 +236,7 @@ void ProteinXtp::parseAccession2dbxref() {
     }
 
 
-    qDebug() << "ProteinXtp::parseAccession2dbxref end" ;
+    //qDebug() << "ProteinXtp::parseAccession2dbxref end" ;
 
 }
 

src/input/mascot/mascotdatparser.cpp

@@ -65,6 +65,12 @@ void MascotDatParser::parse(QIODevice * in_stream) {
                 parseHeaderLine( mime_parser.getCurrentTextStream().readLine());
             }
         }
+        else if (mime_parser.getCurrentFileName() == "masses") {
+            while(!mime_parser.getCurrentTextStream().atEnd()) {
+                parseMassesLine( mime_parser.getCurrentTextStream().readLine());
+            }
+
+        }
         else if (mime_parser.getCurrentFileName() == "summary") {
             while(!mime_parser.getCurrentTextStream().atEnd()) {
                 parseSummaryLine( mime_parser.getCurrentTextStream().readLine());
@@ -99,11 +105,39 @@ void MascotDatParser::parse(QIODevice * in_stream) {
             }
             saveQuery();
         }
+
     }
 
     mime_parser.close();
     qDebug() << "MascotDatParser::parse end";
 }
+
+void MascotDatParser::parseMassesLine(const QString & masses_line) {
+    qDebug() << __FILE__ << " " << __FUNCTION__<< " " << __LINE__ << " " << masses_line;
+    if (_regexp_header_line.exactMatch(masses_line)) {
+        QStringList header_list = _regexp_header_line.capturedTexts();
+//           C_term=17.002740
+//N_term=1.007825
+//delta1=15.994915,Oxidation (M)
+        if (header_list[1].startsWith("delta")) {
+            QStringList delta_mod_list =  header_list[2].split(",");
+            pappso::pappso_double mass = delta_mod_list[0].toDouble();
+            //_number_of_residues = header_list[5].toUInt();
+        }
+
+//NeutralLoss1=0.000000
+//NeutralLoss1_master=63.998285
+//FixedMod1=57.021464,Carbamidomethyl (C)
+        if (header_list[1].startsWith("FixedMod")) {
+            QStringList fixed_mod_list =  header_list[2].split(",");
+            pappso::pappso_double mass = fixed_mod_list[0].toDouble();
+            //_number_of_residues = header_list[2].toUInt();
+        }
+
+//FixedModResidues1=C
+    }
+}
+
 void MascotDatParser::parseProteinLine(const QString & protein_line) {
     qDebug() << __FILE__ << " " << __FUNCTION__<< " " << __LINE__ << " " << protein_line;
     ProteinXtpSp sp_xtp_protein;
@@ -224,14 +258,14 @@ void MascotDatParser::parsePeptidesLine(const QString & peptide_line) {
                     //q1_p1=0,597.302322,0.997884,2,GAWHK,9,0000000,7.97,0000012000000000000,0,0;"sp|O95006|OR2F2_HUMAN":0:299:303:1
                     int position =  value.indexOf(";\"", 0);
                     QString peptide_string = value.mid(0, position);
-                    qDebug() << "peptide_string=" <<  peptide_string;
+                    //qDebug() << "peptide_string=" <<  peptide_string;
 
                     _current_peptide.peptide_string_list = peptide_string.split(",");
 
 
 
                     QString protein_string = value.mid(position+2);
-                    qDebug() << "protein_string=" <<  protein_string;
+                    //qDebug() << "protein_string=" <<  protein_string;
                     //"sp|Q9Y2I7|FYV1_HUMAN":0:670:675:2,"tr|E9PDH4|E9PDH4_HUMAN":0:614:619:2
                     _current_peptide.protein_string_list = protein_string.split(",\"");
                 }
@@ -266,6 +300,7 @@ void MascotDatParser::parsePeptidesLine(const QString & peptide_line) {
             q856_p9_terms=-,L:-,V
             q856_p9_subst=1,X,W
             */
+
         }
     }
 
@@ -415,6 +450,20 @@ void MascotDatParser::savePeptideList(std::vector< PeptideLine> & peptide_list,
     //parse and save
     for(PeptideLine & peptide_line:peptide_list) {
 
+        /*
+         *
+        0 setMissedCleavages(Integer.parseInt(st.nextToken()));
+        1 setPeptideMr(Double.parseDouble(st.nextToken()));
+        2 setDeltaMass(Double.parseDouble(st.nextToken()));
+        3 setNumberOfIonsMatched(Integer.parseInt(st.nextToken()));
+        4 setSequence(st.nextToken());
+        5 setPeaksUsedFromIons1(Integer.parseInt(st.nextToken()));
+        6 setVariableModificationsArray(st.nextToken());
+        7 setIonsScore(Double.parseDouble(st.nextToken()));
+        8 parseIonSeries(st.nextToken());
+        9 setPeaksUsedFromIons2(Integer.parseInt(st.nextToken()));
+        10 setPeaksUsedFromIons3(Integer.parseInt(st.nextToken()));
+        */
         pappso::pappso_double ion_score = peptide_line.peptide_string_list.at(7).toDouble();
         peptide_evidence.setParam(PeptideEvidenceParam::mascot_score,  QVariant(ion_score));
 
@@ -439,6 +488,9 @@ void MascotDatParser::savePeptideList(std::vector< PeptideLine> & peptide_list,
         PeptideXtpSp peptide_sp;
         peptide_sp = PeptideXtp(peptide_str).makePeptideXtpSp();
 
+        //variable modifications :
+        setVariableModifications(peptide_sp, peptide_line.peptide_string_list.at(6));
+
         peptide_evidence.setPeptideXtpSp(_p_project->getPeptideStore().getInstance(peptide_sp));
 
 
@@ -456,7 +508,7 @@ void MascotDatParser::savePeptideList(std::vector< PeptideLine> & peptide_list,
             if (is_decoy) {
                 accession = QString("%1|reversed").arg(accession);
             }
-            qDebug() << "accession=" <<  accession;
+            //qDebug() << "accession=" <<  accession;
             QStringList position_list = str.mid(position+2).split(":");
             if (position_list.size() != 4) {
                 throw pappso::PappsoException(QObject::tr("ERROR position_list.size() != 4 %1").arg(str));
@@ -501,3 +553,6 @@ pappso::pappso_double MascotDatParser::getEvalueExperimental(pappso::pappso_doub
     pappso::pappso_double eval = (ion_score - summary_line.plug_hole) / (pappso::pappso_double)10;
     eval = confidence_interval / pow((pappso::pappso_double)10, eval);
 }
+
+void MascotDatParser::setVariableModifications(PeptideXtpSp & peptide_sp, QString variable_modifications_string) {
+}

src/input/mascot/mascotdatparser.h

@@ -45,8 +45,10 @@ private:
     void parsePeptidesLine(const QString & peptide_line);
     void parseQueryLine(const QString & query_line);
     void parseSummaryLine(const QString & summary_line);
+    void parseMassesLine(const QString & masses_line);
     void saveQuery();
     void saveAndClearPeptide();
+    void setVariableModifications(PeptideXtpSp & peptide_sp, QString variable_modifications_string);