WIP: Mascot parser

src/input/mascot/mascotdatparser.cpp

@@ -147,6 +147,39 @@ void MascotDatParser::parseHeaderLine(const QString & header_line) {
     }
 }
 
+void MascotDatParser::saveAndClearPeptide() {
+    qDebug() << "MascotDatParser::saveAndClearPeptide begin";
+    if (_current_peptide.query_index > 0) {
+        //parse and save
+
+        pappso::Peptide peptide(_current_peptide.peptide_string_list.at(4));
+
+        if (_current_peptide.protein_string_list.size() != _current_peptide.fasta_file_list.size()) {
+            throw pappso::PappsoException(QObject::tr("ERROR (_current_peptide.protein_string_list.size() != _current_peptide.fasta_file_list.size()) %1").arg(_current_peptide.protein_string_list.join(",\"")));
+        }
+
+        foreach (const QString &str, _current_peptide.protein_string_list) {
+            //sp|O95006|OR2F2_HUMAN":0:299:303:1
+            int position =  str.indexOf("\"", 0);
+            QString accession = str.mid(0, position);
+            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));
+            }
+            unsigned int start = position_list.at(1).toUInt();
+            unsigned int stop = position_list.at(2).toUInt();
+        }
+    }
+
+    //new peptide query clear
+    _current_peptide.peptide_string_list.clear();
+    _current_peptide.fasta_file_list.clear();
+    _current_peptide.query_index = 0;
+    _current_peptide.subst = "";
+    qDebug() << "MascotDatParser::saveAndClearPeptide end";
+}
+
 void MascotDatParser::parsePeptidesLine(const QString & peptide_line) {
     try {
         if (_regexp_header_line.exactMatch(peptide_line)) {
@@ -156,14 +189,18 @@ void MascotDatParser::parsePeptidesLine(const QString & peptide_line) {
             QStringList index_list = index.split("_");
             if (index_list.size() == 3) {
                 if (index_list[2] == "db") {
+                    saveAndClearPeptide();
                     //q1_p1_db=02
-                    _peptides_fasta_file_list.clear();
                     while (value.size() > 0) {
                         QString fasta_str = value.mid(0,2);
-                        _peptides_fasta_file_list.push_back(_fasta_file_list.at(fasta_str.toInt()-1));
+                        _current_peptide.fasta_file_list.push_back(_fasta_file_list.at(fasta_str.toInt()-1));
                         value = value.mid(2);
                     }
                 }
+                //q856_p9_subst=1,X,W
+                else if (index_list[2] == "subst") {
+                    _current_peptide.subst = value;
+                }
             }
             else if (index_list.size() == 2) {
                 if (value == "-1") {
@@ -172,37 +209,22 @@ void MascotDatParser::parsePeptidesLine(const QString & peptide_line) {
                 else {
 
                     QString query_index = index_list[0];
+                    _current_peptide.query_index = query_index.mid(1).toUInt();
                     QString peptide_index = index_list[1];
+                    _current_peptide.peptide_index = peptide_index.mid(1).toUInt();
                     //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;
 
-                    QStringList peptide_string_list = peptide_string.split(",");
-                    pappso::Peptide peptide(peptide_string_list.at(4));
+                    _current_peptide.peptide_string_list = peptide_string.split(",");
 
 
 
                     QString protein_string = value.mid(position+2);
                     qDebug() << "protein_string=" <<  protein_string;
                     //"sp|Q9Y2I7|FYV1_HUMAN":0:670:675:2,"tr|E9PDH4|E9PDH4_HUMAN":0:614:619:2
-                    QStringList protein_string_list = protein_string.split(",\"");
-                    if (protein_string_list.size() != _peptides_fasta_file_list.size()) {
-                        throw pappso::PappsoException(QObject::tr("ERROR (protein_string_list.size() != _peptides_fasta_file_list.size()) %1").arg(value));
-                    }
-
-                    foreach (const QString &str, protein_string_list) {
-                        //sp|O95006|OR2F2_HUMAN":0:299:303:1
-                        int position =  str.indexOf("\"", 0);
-                        QString accession = str.mid(0, position);
-                        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(value));
-                        }
-                        unsigned int start = position_list.at(1).toUInt();
-                        unsigned int stop = position_list.at(2).toUInt();
-                    }
+                    _current_peptide.protein_string_list = protein_string.split(",\"");
                 }
 
 

src/input/mascot/mascotdatparser.h

@@ -43,6 +43,16 @@ private:
     void parseProteinLine(const QString & protein_line);
     void parseHeaderLine(const QString & protein_line);
     void parsePeptidesLine(const QString & peptide_line);
+    void saveAndClearPeptide();
+    
+    struct PeptideLine {
+        unsigned int query_index=0;
+        unsigned int peptide_index=0;
+        QString subst;
+        QStringList peptide_string_list;
+        QStringList protein_string_list;
+        std::vector<FastaFileSp> fasta_file_list;
+    };
 private:
     Project * _p_project;
     IdentificationGroup * _p_identification_group;
@@ -52,13 +62,16 @@ private:
     ProteinXtp _current_protein;
     std::vector<FastaFileSp> _fasta_file_list;
 
-    std::vector<FastaFileSp> _peptides_fasta_file_list;
+    
 
     QRegExp   _regexp_header_line;
     unsigned int _number_of_queries=0;
     unsigned int _number_of_residues=0;
     QString _error_str;
     
+    PeptideLine _current_peptide;
+    
+    
 };
 
 #endif // MASCOTDATPARSER_H