parsing accession to find dbxref list

src/core/proteinxtp.cpp

@@ -76,9 +76,9 @@ QString ProteinXtp::getOnlyAminoAcidSequence() const {
 
 pappso::pappso_double ProteinXtp::getMass() const {
     try {
-    pappso::Peptide peptide(getOnlyAminoAcidSequence().replace("X",""));
-    return peptide.getMass();
-    
+        pappso::Peptide peptide(getOnlyAminoAcidSequence().replace("X",""));
+        return peptide.getMass();
+
     }
     catch (pappso::PappsoException error) {
         throw pappso::PappsoException(QObject::tr("Error computing mass for protein %1 :\n%2").arg(getAccession()).arg(error.qwhat()));
@@ -97,27 +97,64 @@ public:
 unsigned int ProteinXtp::countTrypticPeptidesForPAI() const {
     qDebug() << "ProteinXtp::countTrypticPeptidesForPAI begin";
     try {
-    pappso::Enzyme kinase;
-    kinase.setMiscleavage(0);
-    DigestionHandler digestion;
-
-    pappso::ProteinSp protein = std::make_shared<const pappso::Protein>(this->getDescription(),this->getOnlyAminoAcidSequence().replace("X",""));
-    kinase.eat(0,protein,false,digestion);
-
-    unsigned int count = 0;
-    for (const QString & peptide_str: digestion._peptide_list) {
-        pappso::Peptide peptide(peptide_str);
-        pappso::mz mass= peptide.getMass();
-        if ((mass > 800) && (mass < 2500)) {
-            count ++;
+        pappso::Enzyme kinase;
+        kinase.setMiscleavage(0);
+        DigestionHandler digestion;
+
+        pappso::ProteinSp protein = std::make_shared<const pappso::Protein>(this->getDescription(),this->getOnlyAminoAcidSequence().replace("X",""));
+        kinase.eat(0,protein,false,digestion);
+
+        unsigned int count = 0;
+        for (const QString & peptide_str: digestion._peptide_list) {
+            pappso::Peptide peptide(peptide_str);
+            pappso::mz mass= peptide.getMass();
+            if ((mass > 800) && (mass < 2500)) {
+                count ++;
+            }
         }
-    }
 
-    qDebug() << "ProteinXtp::countTrypticPeptidesForPAI end";
-    return count;
-    
+        qDebug() << "ProteinXtp::countTrypticPeptidesForPAI end";
+        return count;
+
     }
     catch (pappso::PappsoException error) {
         throw pappso::PappsoException(QObject::tr("Error in countTrypticPeptidesForPAI for protein %1 :\n%2").arg(getAccession()).arg(error.qwhat()));
     }
 }
+
+
+void ProteinXtp::parseAccession2dbxref() {
+    QStringList access_list = getAccession().split("\\|");
+    // if (access.length == 1)
+    QRegExp atg("^[Aa][Tt][MmCc1-5][Gg]\\d{5}\\.?\\d?$");
+    QRegExp ncbi_gi("^[0-9]{5,8}$");
+    QRegExp swiss_prot("^P[A-Z0-9]{5}$");
+    QRegExp trembl("^[QOA][A-Z0-9]{5}$");
+    QRegExp ref ("^[NZ]P\\_[0-9]{5,8}$");
+
+    for (QString & accession :access_list) {
+        if (atg.indexIn(accession, 0) != -1) {
+            QStringList temp = accession.split("\\.");
+            _dbxref_list.push_back(std::make_pair(ExternalDatabase::AGI_LocusCode, temp.at(0)));
+        }
+        if (ncbi_gi.indexIn(accession, 0) != -1) {
+            _dbxref_list.push_back(std::make_pair(ExternalDatabase::NCBI_gi, accession));
+        }
+        
+        if (swiss_prot.indexIn(accession, 0) != -1) {
+            _dbxref_list.push_back(std::make_pair(ExternalDatabase::SwissProt, accession));
+        }
+        
+        if (trembl.indexIn(accession, 0) != -1) {
+            _dbxref_list.push_back(std::make_pair(ExternalDatabase::TrEMBL, accession));
+        }
+        if (ref.indexIn(accession, 0) != -1) {
+            _dbxref_list.push_back(std::make_pair(ExternalDatabase::ref, accession));
+        }
+    }
+
+}
+
+const std::list<std::pair<ExternalDatabase, QString>> & ProteinXtp::getDbxrefList() const {
+    return _dbxref_list;
+}

src/core/proteinxtp.h

@@ -23,7 +23,9 @@
 
 #include <pappsomspp/protein/protein.h>
 #include <pappsomspp/types.h>
+#include <list>
 #include "sequencedatabase.h"
+#include "../utils/types.h"
 
 #ifndef PROTEIN_XTP_H
 #define PROTEIN_XTP_H
@@ -61,11 +63,18 @@ public:
     /** @brief get protein mass in dalton
      */
     pappso::pappso_double getMass() const;
+    
+    /** @brief look for external database accessions in accesion text
+     */
+    void parseAccession2dbxref();
+    
+    const std::list<std::pair<ExternalDatabase, QString>> & getDbxrefList() const;
 
 private:
     SequenceDatabase * _p_sequence_database;
     bool _is_decoy=false;
     bool _is_contaminant=false;
+    std::list<std::pair<ExternalDatabase, QString>> _dbxref_list;
 };
 
 #endif // PROTEIN_XTP_H

src/output/proticdbml.h

@@ -51,6 +51,8 @@ private :
     void writeOboModif(pappso::AaModificationP mod);
     void writeCvParam(QString acc, QString value, QString description);
     void writeSequence(ProteinMatch * p_protein_match);
+    void writeProject();
+    void writeIdentMethod();
     
 private :
     QFile * _output_file;

src/utils/proteinstore.cpp

@@ -88,6 +88,7 @@ void ProteinStore::setProteinInformations(ProteinXtpSp & peptide_in) {
     peptide_in.get()->setIsContaminant(false);
     peptide_in.get()->setIsDecoy(false);
     QString accession = peptide_in.get()->getAccession();
+    peptide_in.get()->parseAccession2dbxref();
     
     if ((!_regexp_contaminant.isEmpty()) && (_regexp_contaminant.indexIn(accession, 0)>-1)) {
         //qDebug() << "ProteinStore::setProteinInformations is contaminant " << accession;

src/utils/types.h

@@ -28,6 +28,17 @@
 #include <cstdint>
 /*********** enumerations *********************************/
 
+/** \def ExternalDatabase external database references
+ *
+ */
+enum class ExternalDatabase {
+    AGI_LocusCode, ///< AGI_LocusCode
+    NCBI_gi, ///< NCBI_gi
+    SwissProt, ///< Swiss-Prot
+    TrEMBL, ///< TrEMBL
+    ref ///< ref
+};
+
 /** \def IdentificationEngine identification engine
  *
  */

src/utils/utils.cpp

 #include "utils.h"
 
 const QUrl Utils::getOlsUrl(QString psimod_accession) {
-    
+
     QString iri(QString("http://purl.obolibrary.org/obo/%1").arg(psimod_accession.replace(":","_")));
     QUrl url(QString("http://www.ebi.ac.uk/ols/ontologies/mod/terms?iri=%1").arg(iri));
     return url;
 }
+
+
+const QString Utils::getDatabaseName(ExternalDatabase database) {
+    QString database_name;
+    switch (database) {
+    case ExternalDatabase::AGI_LocusCode :
+        database_name = "AGI_LocusCode";
+        break;
+    case ExternalDatabase::NCBI_gi :
+        database_name = "NCBI_gi";
+        break;
+
+    case ExternalDatabase::SwissProt :
+        database_name = "Swiss-Prot";
+        break;
+    case ExternalDatabase::TrEMBL :
+        database_name = "TrEMBL";
+        break;
+    case ExternalDatabase::ref :
+        database_name = "ref";
+        break;
+    }
+    return database_name;
+}

src/utils/utils.h

@@ -22,6 +22,7 @@
 ******************************************************************************/
 #include<QUrl>
 #include <QString>
+#include "types.h"
 
 #ifndef UTILS_H
 #define UTILS_H
@@ -29,6 +30,7 @@ class Utils
 {
 public:
     static const QUrl getOlsUrl(QString modification);
+    static const QString getDatabaseName(ExternalDatabase database);
 };
 
 #endif // UTILS_H