From 4fe5dbdbc725448b63a3eea7135a3d7d230cb3c5 Mon Sep 17 00:00:00 2001
From: Olivier Langella <Olivier.Langella@moulon.inra.fr>
Date: Tue, 16 May 2017 16:27:01 +0200
Subject: [PATCH] parsing accession to find dbxref list
---
src/core/proteinxtp.cpp | 77 +-
src/core/proteinxtp.h | 9 +
src/output/proticdbml.cpp | 1370 ++++++++++++++++++------------------
src/output/proticdbml.h | 2 +
src/utils/proteinstore.cpp | 1 +
src/utils/types.h | 11 +
src/utils/utils.cpp | 26 +-
src/utils/utils.h | 2 +
8 files changed, 787 insertions(+), 711 deletions(-)
diff --git a/src/core/proteinxtp.cpp b/src/core/proteinxtp.cpp
index 3f468ea4..ef3bba3f 100644
--- a/src/core/proteinxtp.cpp
+++ b/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;
+}
diff --git a/src/core/proteinxtp.h b/src/core/proteinxtp.h
index 9356fdc6..947ebb59 100644
--- a/src/core/proteinxtp.h
+++ b/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
diff --git a/src/output/proticdbml.cpp b/src/output/proticdbml.cpp
index ff313956..1ace5b09 100644
--- a/src/output/proticdbml.cpp
+++ b/src/output/proticdbml.cpp
@@ -77,702 +77,692 @@ void ProticdbMl::write(ProjectSp sp_project) {
if (_sp_project.get() == nullptr) {
throw pappso::PappsoException(QObject::tr("Error writing PROTICdbML file :\n project is empty"));
}
-
-
+
+
_output_stream->writeNamespace("http://www.w3.org/2001/XMLSchema-instance","xsi");
- // writer.setDefaultNamespace(namespaceURI);
- _output_stream->writeStartElement("PROTICdb");
- _output_stream->writeAttribute("xmlns","http://pappso.inra.fr/xsd/masschroqml/2.2");
- _output_stream->writeAttribute("http://www.w3.org/2001/XMLSchema-instance","schemaLocation","http://pappso.inra.fr/xsd/masschroqml/2.2 http://pappso.inra.fr/xsd/masschroq-2.2.xsd");
-
+ // writer.setDefaultNamespace(namespaceURI);
+ _output_stream->writeStartElement("PROTICdb");
+ //_output_stream->writeAttribute("xmlns","http://pappso.inra.fr/xsd/masschroqml/2.2");
+ //_output_stream->writeAttribute("http://www.w3.org/2001/XMLSchema-instance","schemaLocation","http://pappso.inra.fr/xsd/masschroqml/2.2 http://pappso.inra.fr/xsd/masschroq-2.2.xsd");
+
+
+ //writer.writeAttribute(xmlnsxsi, "noNamespaceSchemaLocation", xsischemaLocation);
- //writer.writeAttribute(xmlnsxsi, "noNamespaceSchemaLocation", xsischemaLocation);
+ _output_stream->writeAttribute("version", "1.0");
- _output_stream->writeAttribute("version", "1.0");
+ _output_stream->writeAttribute("type", "MSidentificationResults");
- _output_stream->writeAttribute("type", "MSidentificationResults");
+ writeIdentMethod();
+ _output_stream->writeStartElement("sequences");
+ for (IdentificationGroup * p_ident_group : sp_project.get()->getIdentificationGroupList()) {
- //writeIdentMethod();
- _output_stream->writeStartElement("sequences");
-for (IdentificationGroup * p_ident_group : sp_project.get()->getIdentificationGroupList()) {
-
- for (ProteinMatch * p_protein_match: p_ident_group->getProteinMatchList()) {
- writeSequence(p_protein_match);
+ for (ProteinMatch * p_protein_match: p_ident_group->getProteinMatchList()) {
+ writeSequence(p_protein_match);
+ }
}
+ //</sequences>
+ _output_stream->writeEndElement();
+
+ writeProject();
+
}
- //</sequences>
+
+void ProticdbMl::writeSequence(ProteinMatch * p_protein_match) {
+ qDebug() << "ProticdbMl::writeSequence begin";
+ if (p_protein_match->getValidationState() != ValidationState::grouped) return;
+ pappso::GrpProtein * p_grp_protein = p_protein_match->getGrpProteinSp().get();
+
+ if (_map_accession2xmlid.find(p_grp_protein->getAccession()) == _map_accession2xmlid.end()) {
+ //not found
+
+ QString id = QString("seq%1").arg(_map_accession2xmlid.size());
+ _output_stream->writeStartElement("sequence");
+ _output_stream->writeAttribute("id", id);
+ QString display_id = QString("%1 %2").arg(p_grp_protein->getAccession()).arg(p_protein_match->getProteinXtpSp().get()->getDescription());
+ _output_stream->writeAttribute("display_id", display_id.left(60));
+ // <dbxref key="AT5G16390"
+ // dbname="AGI_LocusCode"></dbxref>
+ //if (prot.get_dbxref_type().equals("no") == false) {
+ for (const std::pair<ExternalDatabase, QString> & dbxref :p_protein_match->getProteinXtpSp().get()->getDbxrefList()) {
+ //if (false) {
+ _output_stream->writeStartElement("dbxref");
+ _output_stream->writeAttribute("dbname",
+ Utils::getDatabaseName (dbxref.first));
+ _output_stream->writeAttribute("key",
+ dbxref.second);
+ _output_stream->writeEndElement();// dbxref
+ }
+ _output_stream->writeStartElement("description");
+ _output_stream->writeCharacters(p_protein_match->getProteinXtpSp().get()->getDescription());
+ _output_stream->writeEndElement();// description
+ _output_stream->writeStartElement("seq");
+ _output_stream->writeCharacters(p_protein_match->getProteinXtpSp().get()->getSequence());
+ _output_stream->writeEndElement();// seq
+
+ _output_stream->writeEndElement();// sequence
+
+ _map_accession2xmlid.insert(std::pair<QString, QString>(p_grp_protein->getAccession(), id));
+
+ }
+ qDebug() << "ProticdbMl::writeSequence end";
+}
+
+void ProticdbMl::writeProject() {
+qDebug() << "ProticdbMl::writeProject begin";
+ _output_stream->writeStartElement("project");
+ _output_stream->writeAttribute("name", "");
+ _output_stream->writeAttribute("id", "p1");
+ // ajout des echantillons_msrun
+ writeSamples();
+ writeMsRuns();
+
+ _output_stream->writeStartElement("identificationRuns");
+
+ for (IdentificationGroup * p_identification : _sp_project.get()->getIdentificationGroupList()) {
+ _output_stream->writeStartElement("identificationRun");
+ // ajout d'une mnip d'identification
+ writeIdentificationRun(p_identification);
+
+ // ajout des petides
+ writepeptideHits(p_identification);
+
+ // ajout des mathces
+ writeMatchs(p_identification);
+
+ _output_stream->writeEndElement();// identificationRun
+ }
+ _output_stream->writeEndElement();// identificationRuns
+ _output_stream->writeEndElement();// project
+
+ _output_stream->writeEndDocument();
+
+
+qDebug() << "ProticdbMl::writeProject end duration = " << duracel << "ms";
+}
+
+void ProticdbMl::writeIdentMethod() {
+ // Ajout des méthodes
+ _output_stream->writeStartElement("identMeth");
+ _output_stream->writeAttribute("name", "");
+ _output_stream->writeAttribute("id", "m1");
+ _output_stream->writeEndElement();
+
+ // Ajout des base de données
+ _output_stream->writeStartElement("customDb");
+ _output_stream->writeAttribute("name", "");
+ _output_stream->writeAttribute("id", "customdb0");
_output_stream->writeEndElement();
- }
-
- void ProticdbMl::writeSequence(ProteinMatch * p_protein_match) {
- qDebug() << "ProticdbMl::writeSequence begin";
- if (p_protein_match->getValidationState() != ValidationState::grouped) return;
- pappso::GrpProtein * p_grp_protein = p_protein_match->getGrpProteinSp().get();
-
- if (_map_accession2xmlid.find(p_grp_protein->getAccession()) == _map_accession2xmlid.end()) {
- //not found
-
- QString id = QString("seq%1").arg(_map_accession2xmlid.size());
- _output_stream->writeStartElement("sequence");
- _output_stream->writeAttribute("id", id);
- QString display_id = QString("%1 %2").arg(p_grp_protein->getAccession()).arg(p_protein_match->getProteinXtpSp().get()->getDescription());
- if (display_id.size() > 60) {
- _output_stream->writeAttribute("display_id", display_id.left(60));
- } else {
- _output_stream->writeAttribute("display_id", display_id);
- }
- // <dbxref key="AT5G16390"
- // dbname="AGI_LocusCode"></dbxref>
- //if (prot.get_dbxref_type().equals("no") == false) {
- if (false) {
- _output_stream->writeStartElement("dbxref");
- _output_stream->writeAttribute("key",
- prot.get_dbxref_accession());
- _output_stream->writeAttribute("dbname",
- prot.get_dbxref_type());
- _output_stream->writeEndElement();// dbxref
- }
- _output_stream->writeStartElement("description");
- _output_stream->writeCharacters(p_protein_match->getProteinXtpSp().get()->getDescription());
- _output_stream->writeEndElement();// description
- _output_stream->writeStartElement("seq");
- _output_stream->writeCharacters(p_protein_match->getProteinXtpSp().get()->getSequence());
- _output_stream->writeEndElement();// seq
-
- _output_stream->writeEndElement();// sequence
-
- _map_accession2xmlid.insert(std::pair<QString, QString>(p_grp_protein->getAccession(), id));
-
+
+
+}
+
+private void writeSamples() throws Exception {
+ try {
+ writer.writeStartElement("samples");
+
+ for (Identification ident : arrayIdentifications) {
+ for (MsRun samp : ident.getMsRunSet()) {
+ writer.writeStartElement("sample");
+ String id_samp = "samp" + sample_to_id.size();
+ String id_msrun = "msr" + msrun_to_id.size();
+ String name = samp.getSampleName();
+ name = name.replaceAll(".xml", "");
+ // balise sample
+ writer.writeAttribute("name", name);
+ writer.writeAttribute("id", id_samp);
+ writer.writeEmptyElement("description");
+ sample_to_id.put(samp.getSampleName(), id_samp);
+
+ // Element spectrumList = doc.createElement("spectrumList");
+ // msRun.appendChild(spectrumList);
+
+ msrun_to_id.put(samp.getSampleName(), id_msrun);
+ writer.writeEndElement();// sample
+ }
+ }
+ writer.writeEndElement();// samples
+ } catch (Exception e) {
+ String message = "error writing samples in ProticDbMl \n"
+ + e.getMessage();
+ logger.error(message);
+ throw new Exception(message);
+ }
+}
+
+private void writeMsRuns() throws Exception {
+ try {
+ writer.writeStartElement("msRuns");
+
+ for (Identification ident : arrayIdentifications) {
+ for (MsRun samp : ident.getMsRunSet()) {
+
+ String name = samp.getSampleName();
+ name = name.replaceAll(".xml", "");
+
+ String id_samp = sample_to_id.get(samp.getSampleName());
+
+ // Element spectrumList = doc.createElement("spectrumList");
+ // msRun.appendChild(spectrumList);
+
+ String id_msrun = msrun_to_id.get(samp.getSampleName());
+
+ // balise MsRun
+ writer.writeStartElement("msRun");
+ writer.writeAttribute("sample_id", id_samp);
+ writer.writeAttribute("id", id_msrun);
+
+ writer.writeStartElement("description");
+ writer.writeStartElement("admin");
+ // writer.writeStartElement("contact");
+ // writer.writeStartElement("email");
+ // writer.writeCharacters("valot@moulon.inra.fr");
+ // writer.writeEndElement();// email
+
+ // writer.writeStartElement("name");
+ // writer.writeCharacters("Valot Benoit");
+ // writer.writeEndElement();// name
+ // writer.writeEndElement();// contact
+
+ writer.writeStartElement("sourceFile");
+ writer.writeStartElement("nameOfFile");
+ writer.writeCharacters(name + ".RAW");
+ writer.writeEndElement();// nameOfFile
+ writer.writeEndElement();// sourceFile
+
+ writer.writeStartElement("sampleName");
+ writer.writeCharacters(name);
+ writer.writeEndElement();// sampleName
+ writer.writeEndElement();// admin
+ writer.writeEndElement();// description
+
+ writer.writeEndElement();// msRun
+ }
+ }
+ writer.writeEndElement();// msRuns
+ } catch (Exception e) {
+ String message = "error writing MS runs in ProticDbMl \n"
+ + e.getMessage();
+ logger.error(message);
+ throw new Exception(message);
+ }
+}
+
+
+private void writeIdentificationRun(Identification identification)
+throws Exception {
+ try {
+ writer.writeAttribute("ident_method_id", "m1");
+ writer.writeAttribute("customdb_id", "customdb0");
+
+ writer.writeStartElement("description");
+ writer.writeStartElement("admin");
+ // writer.writeStartElement("contact");
+ // writer.writeStartElement("email");
+ // writer.writeCharacters("valot@moulon.inra.fr");
+ // writer.writeEndElement();// email
+
+ // writer.writeStartElement("name");
+ // writer.writeCharacters("Valot Benoit");
+ // writer.writeEndElement();// name
+ // writer.writeEndElement();// contact
+ this.xpipFile = XtandemPipelineSession.getInstance()
+ .getCurrentXpipFile();
+
+ writer.writeStartElement("sourceFile");
+ writer.writeStartElement("nameOfFile");
+ // TODO
+ // get the loaded xpip file name
+ writer.writeCharacters(this.xpipFile.getName());
+ writer.writeEndElement();// nameOfFile
+ writer.writeStartElement("pathToFile");
+ writer.writeCharacters(this.xpipFile.getAbsolutePath());
+ writer.writeEndElement();// pathToFile
+ writer.writeStartElement("fileType");
+ writer.writeCharacters("XPIP file");
+ writer.writeEndElement();// fileType
+ writer.writeEndElement();// sourceFile
+
+ writer.writeEndElement();// admin
+
+ // if we can retrieve original informations in xml xtandem results
+ HashSet<IdentificationXtandemFile> xtSourceFiles = Utils
+ .getXtandemFileList(identification);
+ if (xtSourceFiles != null) {
+ //
+ for (IdentificationXtandemFile xtFile : xtSourceFiles) {
+ xtFile.writeProticDbMlDataProcessing(this.writer);
+ }
+ } else {
+ writer.writeStartElement("dataProcessing");
+ writer.writeStartElement("software");
+ writer.writeStartElement("name");
+ writer.writeAttribute("acc", "PROTICdbO:0000283");
+ writer.writeCharacters("X!Tandem");
+ writer.writeEndElement();// name
+ writer.writeStartElement("version");
+ writer.writeCharacters(XtandemPipelineSession.getInstance()
+ .getConfig().getXtandemVersion());
+ writer.writeEndElement();// version
+ writer.writeEndElement();// software
+ writer.writeStartElement("processingMethod");
+ writer.writeEndElement();// processingMethod
+ writer.writeEndElement();// dataProcessing
+ }
+
+ // id: PROTICdbO:0000316
+ // name: X!TandemPipeline
+
+ writer.writeStartElement("dataProcessing");
+ writer.writeStartElement("software");
+ writer.writeStartElement("name");
+ writer.writeAttribute("acc", "PROTICdbO:0000316");
+ writer.writeCharacters("X!TandemPipeline");
+ writer.writeEndElement();// name
+ writer.writeStartElement("version");
+ writer.writeCharacters(XtandemPipelineMain.version);
+ writer.writeEndElement();// version
+ writer.writeEndElement();// software
+ writer.writeStartElement("processingMethod");
+ // cvParams
+
+ this.writeCvParam("PROTICdbO:0000323", ""
+ + XtandemPipelineSession.getInstance().getConfig()
+ .get_protein_evalue(),
+ "X!TandemPipeline filter on protein evalue (log)");
+
+ // indi
+ // combine
+ // phospho
+ xtpExperimentType = XtandemPipelineSession.getInstance()
+ .getDataTypeBase().getType();
+ if (xtpExperimentType.equals("indiv")) {
+ this.writeCvParam("PROTICdbO:0000319", xtpExperimentType, "");
+ }
+ if (xtpExperimentType.equals("combi")) {
+ this.writeCvParam("PROTICdbO:0000320", xtpExperimentType, "");
}
- qDebug() << "ProticdbMl::writeSequence end";
- }
-
- public void close() throws Exception {
- writer.writeEndElement();// sequences
-
- writer.writeStartElement("project");
- writer.writeAttribute("name", "");
- writer.writeAttribute("id", "p1");
- // ajout des echantillons_msrun
- this.writeSamples();
- this.writeMsRuns();
-
- writer.writeStartElement("identificationRuns");
-
- for (Identification identification : this.arrayIdentifications) {
- writer.writeStartElement("identificationRun");
- // ajout d'une mnip d'identification
- this.writeIdentificationRun(identification);
-
- // ajout des petides
- this.writepeptideHits(identification);
-
- // ajout des mathces
- this.writeMatchs(identification);
-
- writer.writeEndElement();// identificationRun
- }
- writer.writeEndElement();// identificationRuns
- writer.writeEndElement();// project
-
- writer.writeEndDocument();
-
- writer.flush();
- writer.close();
-
- logger.info("Duration creating PROTICdbML xml document: "
- + (System.currentTimeMillis() - duracel) + " ms");
- }
-
- private void writeIdentMethod() {
- logger.debug("write begin");
- try {
- // Ajout des méthodes
- writer.writeStartElement("identMeth");
- writer.writeAttribute("name", "");
- writer.writeAttribute("id", "m1");
- writer.writeEndElement();
-
- // Ajout des base de données
- writer.writeStartElement("customDb");
- writer.writeAttribute("name", "");
- writer.writeAttribute("id", "customdb0");
- writer.writeEndElement();
-
- } catch (Exception e) {
- logger.error(e.getMessage());
- }
- logger.debug("write end");
-
- }
-
- private void writeSamples() throws Exception {
- try {
- writer.writeStartElement("samples");
-
- for (Identification ident : arrayIdentifications) {
- for (MsRun samp : ident.getMsRunSet()) {
- writer.writeStartElement("sample");
- String id_samp = "samp" + sample_to_id.size();
- String id_msrun = "msr" + msrun_to_id.size();
- String name = samp.getSampleName();
- name = name.replaceAll(".xml", "");
- // balise sample
- writer.writeAttribute("name", name);
- writer.writeAttribute("id", id_samp);
- writer.writeEmptyElement("description");
- sample_to_id.put(samp.getSampleName(), id_samp);
-
- // Element spectrumList = doc.createElement("spectrumList");
- // msRun.appendChild(spectrumList);
-
- msrun_to_id.put(samp.getSampleName(), id_msrun);
- writer.writeEndElement();// sample
- }
- }
- writer.writeEndElement();// samples
- } catch (Exception e) {
- String message = "error writing samples in ProticDbMl \n"
- + e.getMessage();
- logger.error(message);
- throw new Exception(message);
- }
- }
-
- private void writeMsRuns() throws Exception {
- try {
- writer.writeStartElement("msRuns");
-
- for (Identification ident : arrayIdentifications) {
- for (MsRun samp : ident.getMsRunSet()) {
-
- String name = samp.getSampleName();
- name = name.replaceAll(".xml", "");
-
- String id_samp = sample_to_id.get(samp.getSampleName());
-
- // Element spectrumList = doc.createElement("spectrumList");
- // msRun.appendChild(spectrumList);
-
- String id_msrun = msrun_to_id.get(samp.getSampleName());
-
- // balise MsRun
- writer.writeStartElement("msRun");
- writer.writeAttribute("sample_id", id_samp);
- writer.writeAttribute("id", id_msrun);
-
- writer.writeStartElement("description");
- writer.writeStartElement("admin");
- // writer.writeStartElement("contact");
- // writer.writeStartElement("email");
- // writer.writeCharacters("valot@moulon.inra.fr");
- // writer.writeEndElement();// email
-
- // writer.writeStartElement("name");
- // writer.writeCharacters("Valot Benoit");
- // writer.writeEndElement();// name
- // writer.writeEndElement();// contact
-
- writer.writeStartElement("sourceFile");
- writer.writeStartElement("nameOfFile");
- writer.writeCharacters(name + ".RAW");
- writer.writeEndElement();// nameOfFile
- writer.writeEndElement();// sourceFile
-
- writer.writeStartElement("sampleName");
- writer.writeCharacters(name);
- writer.writeEndElement();// sampleName
- writer.writeEndElement();// admin
- writer.writeEndElement();// description
-
- writer.writeEndElement();// msRun
- }
- }
- writer.writeEndElement();// msRuns
- } catch (Exception e) {
- String message = "error writing MS runs in ProticDbMl \n"
- + e.getMessage();
- logger.error(message);
- throw new Exception(message);
- }
- }
-
-
- private void writeIdentificationRun(Identification identification)
- throws Exception {
- try {
- writer.writeAttribute("ident_method_id", "m1");
- writer.writeAttribute("customdb_id", "customdb0");
-
- writer.writeStartElement("description");
- writer.writeStartElement("admin");
- // writer.writeStartElement("contact");
- // writer.writeStartElement("email");
- // writer.writeCharacters("valot@moulon.inra.fr");
- // writer.writeEndElement();// email
-
- // writer.writeStartElement("name");
- // writer.writeCharacters("Valot Benoit");
- // writer.writeEndElement();// name
- // writer.writeEndElement();// contact
- this.xpipFile = XtandemPipelineSession.getInstance()
- .getCurrentXpipFile();
-
- writer.writeStartElement("sourceFile");
- writer.writeStartElement("nameOfFile");
- // TODO
- // get the loaded xpip file name
- writer.writeCharacters(this.xpipFile.getName());
- writer.writeEndElement();// nameOfFile
- writer.writeStartElement("pathToFile");
- writer.writeCharacters(this.xpipFile.getAbsolutePath());
- writer.writeEndElement();// pathToFile
- writer.writeStartElement("fileType");
- writer.writeCharacters("XPIP file");
- writer.writeEndElement();// fileType
- writer.writeEndElement();// sourceFile
-
- writer.writeEndElement();// admin
-
- // if we can retrieve original informations in xml xtandem results
- HashSet<IdentificationXtandemFile> xtSourceFiles = Utils
- .getXtandemFileList(identification);
- if (xtSourceFiles != null) {
- //
- for (IdentificationXtandemFile xtFile : xtSourceFiles) {
- xtFile.writeProticDbMlDataProcessing(this.writer);
- }
- } else {
- writer.writeStartElement("dataProcessing");
- writer.writeStartElement("software");
- writer.writeStartElement("name");
- writer.writeAttribute("acc", "PROTICdbO:0000283");
- writer.writeCharacters("X!Tandem");
- writer.writeEndElement();// name
- writer.writeStartElement("version");
- writer.writeCharacters(XtandemPipelineSession.getInstance()
- .getConfig().getXtandemVersion());
- writer.writeEndElement();// version
- writer.writeEndElement();// software
- writer.writeStartElement("processingMethod");
- writer.writeEndElement();// processingMethod
- writer.writeEndElement();// dataProcessing
- }
-
- // id: PROTICdbO:0000316
- // name: X!TandemPipeline
-
- writer.writeStartElement("dataProcessing");
- writer.writeStartElement("software");
- writer.writeStartElement("name");
- writer.writeAttribute("acc", "PROTICdbO:0000316");
- writer.writeCharacters("X!TandemPipeline");
- writer.writeEndElement();// name
- writer.writeStartElement("version");
- writer.writeCharacters(XtandemPipelineMain.version);
- writer.writeEndElement();// version
- writer.writeEndElement();// software
- writer.writeStartElement("processingMethod");
- // cvParams
-
- this.writeCvParam("PROTICdbO:0000323", ""
- + XtandemPipelineSession.getInstance().getConfig()
- .get_protein_evalue(),
- "X!TandemPipeline filter on protein evalue (log)");
-
- // indi
- // combine
- // phospho
- xtpExperimentType = XtandemPipelineSession.getInstance()
- .getDataTypeBase().getType();
- if (xtpExperimentType.equals("indiv")) {
- this.writeCvParam("PROTICdbO:0000319", xtpExperimentType, "");
- }
- if (xtpExperimentType.equals("combi")) {
- this.writeCvParam("PROTICdbO:0000320", xtpExperimentType, "");
- }
- if (xtpExperimentType.equals("phospho")) {
- this.writeCvParam("PROTICdbO:0000321", xtpExperimentType, "");
- }
-
- this.writeCvParam("PROTICdbO:0000325", ""
- + XtandemPipelineSession.getInstance().getConfig()
- .get_peptide_number(),
- "X!TandemPipeline filter on minimal number of peptide per protein");
-
- this.writeCvParam("PROTICdbO:0000324", ""
- + XtandemPipelineSession.getInstance().getConfig()
- .get_peptide_evalue(),
- "X!TandemPipeline filter on peptide evalue");
-
- // TODO write database filter
- // this.writeCvParam("PROTICdbO:0000324", ""
- // +
- // XtandemPipelineSession.getInstance().getConfig().get_database_filter(),
- // "X!TandemPipeline filter on peptide evalue");
-
- writer.writeEndElement();// processingMethod
-
- writer.writeEndElement();// dataProcessing
-
- writer.writeEndElement();// description
- } catch (Exception e) {
- String message = "error writing identificationRun in ProticDbMl \n"
- + e.getMessage();
- logger.error(message);
- throw new Exception(message);
- }
- }
-
- private void writePeptideHit(Peptide pep, Group group,
- PeptideMass peptideMass) throws XMLStreamException, MSMSException {
-
- if (peptideMass == null) {
- String message = "peptideMass == null";
- logger.error(message);
- throw new MSMSException(message);
- }
-
- String pep_hit_id = "pep" + pepSampleScanCle2xmlId.size();
- // param par default
- writer.writeStartElement("peptideHit");
- writer.writeAttribute("calc_mr", "" + (pep.get_mhplus_theo() - 1.00794));
- double exp_mz = (pep.get_mhplus_obser() + (1.00794 * (pep.getCharge() - 1)))
- / pep.getCharge();
- writer.writeAttribute("exp_mz", "" + exp_mz);
- writer.writeAttribute("delta", "" + pep.get_deltamass());
- writer.writeAttribute("exp_mr", "" + (pep.get_mhplus_obser() - 1.00794));
- writer.writeAttribute("acq_number", "" + pep.get_scan());
- writer.writeAttribute("ms_run_id",
- "" + msrun_to_id.get(pep.get_sample()));
- writer.writeAttribute("id", pep_hit_id);
- writer.writeAttribute("exp_z", "" + pep.getCharge());
-
- this.writeCvParam("PROTICdbO:0000339",
- Utils.getPappsoPeptideMassId(group, peptideMass),
- "peptide mass id");
-
- if (xtpExperimentType.equals("phospho")) {
- this.writeCvParam("PROTICdbO:0000349",
- Utils.getPappsoPhosphoPeptideMassId(group, peptideMass),
- "phosphopeptide mass id");
- }
-
- // pep.getMsRun().
- IdentificationDataSource idDataSource = pep
- .getIdentificationDataSource();
- if (idDataSource.getClass().equals(IdentificationXtandemFile.class)) {
- // cvparam specifique xtandem
-
- this.writeCvParam("PROTICdbO:0000287", "" + pep.get_evalue(),
- "xtandem peptide evalue");
-
- this.writeCvParam("PROTICdbO:0000288", "" + pep.get_hypercorr(),
- "xtandem peptide hyperscore");
- }
-
- this.writeCvParam("PROTICdbO:0000289",
- "" + pep.getRetentionTimeInMinutes(),
- "xtandem peptide retention time");
- /*
- * writer.writeStartElement("cvParam"); writer.writeAttribute("value",
- * "" + pep.get_RT()); writer.writeAttribute("name",
- * "xtandem peptide retention time"); writer.writeAttribute("cvLabel",
- * "PROTICdbO"); writer.writeAttribute("accession",
- * "PROTICdbO:0000289"); writer.writeEndElement();// cvParam
- */
-
- // Element cvParam4 = doc.createElement("cvParam");
- // cvParam4.setAttribute("value", "" + pep.get_post());
- // cvParam4.setAttribute("name", "xtandem peptide post");
- // cvParam4.setAttribute("cvLabel", "PROTICdbO");
- // cvParam4.setAttribute("accession", "PROTICdbO:00000XX");
- // peptideHit.appendChild(cvParam4);
- //
- // Element cvParam5 = doc.createElement("cvParam");
- // cvParam5.setAttribute("value", "" + pep.get_pre());
- // cvParam5.setAttribute("name", "xtandem peptide pre");
- // cvParam5.setAttribute("cvLabel", "PROTICdbO");
- // cvParam5.setAttribute("accession", "PROTICdbO:00000XX");
- // peptideHit.appendChild(cvParam5);
-
- // sequences avec les modifs
- writer.writeStartElement("pepSeq");
- writer.writeStartElement("peptide");
- writer.writeCharacters("" + pep.getSequence());
- writer.writeEndElement();// peptide
- if (pep.get_Modifs().size() != 0) {
- this.writePtms(pep.get_Modifs());
- }
- writer.writeEndElement();// pepSeq
-
- pepSampleScanCle2xmlId.put(pep.get_sample_scan_sequencepeptide_cle(),
- pep_hit_id);
- writer.writeEndElement();// peptideHit
- }
-
- private void writePtms(ArrayList<Modifs> mods) throws XMLStreamException {
- writer.writeStartElement("ptms");
- for (Modifs mod : mods) {
- writer.writeStartElement("ptm");
- writer.writeAttribute("diff_mono", "" + mod.get_modvalue());
- writer.writeAttribute("position", "" + mod.get_posi());
- writer.writeAttribute("aa", "" + mod.get_AA());
- try {
- this.writeOboModif(mod.getPsiMod());
- } catch (MSMSException e) {
- String message = "error writing OBO modif : " + e.getMessage();
- logger.warn(message);
- }
- writer.writeEndElement();// ptm
-
- }
- writer.writeEndElement();// ptms
- }
-
- void ProticdbMl::writeOboModif(pappso::AaModificationP mod) {
- _output_stream->writeStartElement("cvParam");
- _output_stream->writeAttribute("name", mod->getName());
- _output_stream->writeAttribute("cvLabel", "MOD");
- _output_stream->writeAttribute("accession", mod->getAccession());
- _output_stream->writeEndElement();// cvParam
- }
-
- private void writepeptideHits(Identification identification)
- throws MSMSException {
- try {
- writer.writeStartElement("peptideHits");
- for (Group group : identification.getGrouping().getGroupList()) {
- // generating peptide numbers
- PeptideMassSet pepMassSet = new PeptideMassSet();
-
- Hashtable<String, PeptideMass> pepSampleScanCle2peptideMass;
- if (this.xtpExperimentType.equals("phospho")) {
- pepSampleScanCle2peptideMass = pepMassSet
- .numberingPhosphoGetSampleScanCle2peptideMass(group);
-
- } else {
- pepSampleScanCle2peptideMass = pepMassSet
- .numberingGetSampleScanCle2peptideMass(group);
- }
-
- for (SubGroup sg : group.getSubGroupSet().getSubGroupList()) {
- for (HashSampleScanSetProt hashProt : sg
- .getHashSampleScanSetProtList()) {
-
- ArrayList<Peptide> peptideList = null;
- if (this.xtpExperimentType.equals("phospho")) {
- peptideList = ((Match) hashProt.getGrpProtMatch())
- .getValidPeptideInPhosphoIsland((HashSampleScanSetPhospho) hashProt);
- } else {
- peptideList = ((Match) hashProt.getGrpProtMatch())
- .get_valid_peptides();
- }
-
- for (Peptide pep : peptideList) {
-
- if (pepSampleScanCle2xmlId.containsKey(pep
- .get_sample_scan_sequencepeptide_cle()) == false) {
-
- writePeptideHit(
- pep,
- group,
- pepSampleScanCle2peptideMass.get(pep
- .get_sample_scan_sequencepeptide_cle()));
-
- }
- }
- }
- }
- }
- writer.writeEndElement();// "peptideHits");
- } catch (Exception e) {
- String message = "error writing peptideHits in ProticDbMl \n"
- + e.getMessage();
- logger.error(message);
- throw new MSMSException(message);
- }
- }
-
- private void writeMatchs(Identification identification)
- throws MSMSException {
- try {
- writer.writeStartElement("matchs");
- for (Group group : identification.getGrouping().getGroupList()) {
- for (SubGroup sg : group.getSubGroupSet().getSubGroupList()) {
- writer.writeStartElement("match");
-
- // echantillons
- Hashtable<String, String> view = new Hashtable<String, String>();
- for (HashSampleScanSetProt hashProt : sg
- .getHashSampleScanSetProtList()) {
- Match match = (Match) hashProt.getGrpProtMatch();
- for (MsRun sample : match.get_sample_matchs()) {
- if (!view.containsKey(sample.getSampleName())) {
- writer.writeStartElement("matchSample");
- writer.writeAttribute(
- "sample_id",
- ""
- + sample_to_id.get(sample
- .getSampleName()));
- String name = sample.getSampleName();
- name = name.replaceAll(".xml", "");
- writer.writeAttribute("name", "" + name);
- view.put(sample.getSampleName(), "ok");
- writer.writeEndElement();// "matchSample");
- }
- }
- }
-
- // proteins
- int k = 1;
- for (HashSampleScanSetProt hashProt : sg
- .getHashSampleScanSetProtList()) {
- writer.writeStartElement("proteinHit");
- Protein prot = (Protein) hashProt.getGrpProtMatch()
- .getGrpProtein();
- writer.writeAttribute("sequence_id",
- "" + prot_to_id.get(prot.get_description()));
- // proteinHit.setAttribute("score", "");
- writer.writeAttribute("rank", "" + k);
-
- // //cvparam
- this.writeCvParam("PROTICdbO:0000284",
- "" + prot.get_mw(),
- "MW determination by Benuch");
- /*
- * writer.writeStartElement("cvParam");
- * writer.writeAttribute("value", "" + prot.get_mw());
- * writer.writeAttribute("name",
- * "MW determination by Benuch");
- * writer.writeAttribute("cvLabel", "PROTICdbO");
- * writer.writeAttribute("accession",
- * "PROTICdbO:0000284"); writer.writeEndElement();//
- * cvParam
- */
- // evalue
- MsRunSet samples = ((Match) hashProt.getGrpProtMatch())
- .get_sample_matchs();
- this.writeCvParam("PROTICdbO:0000291",
- "" + prot.get_evalue(), "Xtandem log evalue");
- /*
- * writer.writeStartElement("cvParam");
- * writer.writeAttribute("value", "" +
- * prot.get_evalue());
- *
- * writer.writeAttribute("name", "Xtandem log evalue");
- * writer.writeAttribute("cvLabel", "PROTICdbO");
- * writer.writeAttribute("accession",
- * "PROTICdbO:0000291"); writer.writeEndElement();//
- * cvParam
- */
-
- // coverage
- if (samples.size() == 1) {
- this.writeCvParam(
- "PROTICdbO:0000285",
- ""
- + ((Match) hashProt
- .getGrpProtMatch())
- .get_coverage(),
- "protein coverage");
- }
-
- // [Term]
- // id: PROTICdbO:0000335
- // name: X!TandemPipeline PAI
- this.writeCvParam("PROTICdbO:0000335",
- "" + hashProt.getPAI(), "PAI");
-
- // [Term]
- // id: PROTICdbO:0000337
- // name: protein group number
- String pappsoGroupingNumber = Utils
- .getPappsoGroupingNumber(group, sg, k);
- if (pappsoGroupingNumber != null) {
- this.writeCvParam("PROTICdbO:0000337",
- pappsoGroupingNumber, "grouping number");
- }
-
- writer.writeEndElement();// "proteinHit");
- k++;
- }
- for (HashSampleScanSetProt hashProt : sg
- .getHashSampleScanSetProtList()) {
- Protein prot2 = (Protein) hashProt.getGrpProtMatch()
- .getGrpProtein();
- // peptidesHitRef
-
- ArrayList<Peptide> peptideList = null;
- if (this.xtpExperimentType.equals("phospho")) {
- peptideList = ((Match) hashProt.getGrpProtMatch())
- .getValidPeptideInPhosphoIsland((HashSampleScanSetPhospho) hashProt);
- } else {
- peptideList = ((Match) hashProt.getGrpProtMatch())
- .get_valid_peptides();
- }
-
- for (Peptide pep : peptideList) {
- // for (Peptide pep : ((Match)
- // hashProt.getGrpProtMatch())
- // .get_peptide_order_inclus_redon()) {
- String xmlId = pepSampleScanCle2xmlId.get(pep
- .get_sample_scan_sequencepeptide_cle());
- if (xmlId == null) {
- String message = "the xml of peptide "
- + pep.get_sample_scan_sequencepeptide_cle()
- + " is not known in pepSampleScanCle2xmlId";
- logger.debug(message);
- throw new Exception(message);
- }
- writer.writeStartElement("peptideHitRef");
- writer.writeAttribute("peptide_hit_id", "" + xmlId);
-
- writer.writeStartElement("fromSeq");
- writer.writeAttribute(
- "seq_id",
- ""
- + prot_to_id.get(prot2
- .get_description()));
- writer.writeAttribute("start", "" + pep.get_start());
- writer.writeAttribute("stop", "" + pep.get_stop());
- String res1 = pep.get_pre();
- res1 = res1.replaceAll("]", "-");
- writer.writeAttribute("residue_before_nter", ""
- + res1.charAt((res1.length() - 1)));
- String res2 = pep.get_post();
- res2 = res2.replaceAll("]", "-");
- writer.writeAttribute("residue_after_cter", ""
- + res2.charAt(0));
-
- writer.writeEndElement();// fromSeq
- writer.writeEndElement();// peptideHitRef
- }
-
- }
- writer.writeEndElement();// "match");
- }
- }
- writer.writeEndElement();// "matchs");
- } catch (Exception e) {
- String message = "error writing Matchs in ProticDbMl \n"
- + e.getMessage();
- logger.error(message);
- throw new MSMSException(message);
- }
- }
-
- void ProticdbMl::writeCvParam(QString acc, QString value, QString description) {
- _output_stream->writeStartElement("cvParam");
-
- _output_stream->writeAttribute("value",value);
- if (description.isEmpty()) {
- _output_stream->writeAttribute("name", "N.A.");
- } else {
- _output_stream->writeAttribute("name", description);
- }
- _output_stream->writeAttribute("cvLabel", "PROTICdbO");
- _output_stream->writeAttribute("accession", acc);
- _output_stream->writeEndElement();// cvParam
-
- }
+ if (xtpExperimentType.equals("phospho")) {
+ this.writeCvParam("PROTICdbO:0000321", xtpExperimentType, "");
+ }
+
+ this.writeCvParam("PROTICdbO:0000325", ""
+ + XtandemPipelineSession.getInstance().getConfig()
+ .get_peptide_number(),
+ "X!TandemPipeline filter on minimal number of peptide per protein");
+
+ this.writeCvParam("PROTICdbO:0000324", ""
+ + XtandemPipelineSession.getInstance().getConfig()
+ .get_peptide_evalue(),
+ "X!TandemPipeline filter on peptide evalue");
+
+ // TODO write database filter
+ // this.writeCvParam("PROTICdbO:0000324", ""
+ // +
+ // XtandemPipelineSession.getInstance().getConfig().get_database_filter(),
+ // "X!TandemPipeline filter on peptide evalue");
+
+ writer.writeEndElement();// processingMethod
+
+ writer.writeEndElement();// dataProcessing
+
+ writer.writeEndElement();// description
+ } catch (Exception e) {
+ String message = "error writing identificationRun in ProticDbMl \n"
+ + e.getMessage();
+ logger.error(message);
+ throw new Exception(message);
+ }
+}
+
+private void writePeptideHit(Peptide pep, Group group,
+ PeptideMass peptideMass) throws XMLStreamException, MSMSException {
+
+ if (peptideMass == null) {
+ String message = "peptideMass == null";
+ logger.error(message);
+ throw new MSMSException(message);
+ }
+
+ String pep_hit_id = "pep" + pepSampleScanCle2xmlId.size();
+ // param par default
+ writer.writeStartElement("peptideHit");
+ writer.writeAttribute("calc_mr", "" + (pep.get_mhplus_theo() - 1.00794));
+ double exp_mz = (pep.get_mhplus_obser() + (1.00794 * (pep.getCharge() - 1)))
+ / pep.getCharge();
+ writer.writeAttribute("exp_mz", "" + exp_mz);
+ writer.writeAttribute("delta", "" + pep.get_deltamass());
+ writer.writeAttribute("exp_mr", "" + (pep.get_mhplus_obser() - 1.00794));
+ writer.writeAttribute("acq_number", "" + pep.get_scan());
+ writer.writeAttribute("ms_run_id",
+ "" + msrun_to_id.get(pep.get_sample()));
+ writer.writeAttribute("id", pep_hit_id);
+ writer.writeAttribute("exp_z", "" + pep.getCharge());
+
+ this.writeCvParam("PROTICdbO:0000339",
+ Utils.getPappsoPeptideMassId(group, peptideMass),
+ "peptide mass id");
+
+ if (xtpExperimentType.equals("phospho")) {
+ this.writeCvParam("PROTICdbO:0000349",
+ Utils.getPappsoPhosphoPeptideMassId(group, peptideMass),
+ "phosphopeptide mass id");
+ }
+
+ // pep.getMsRun().
+ IdentificationDataSource idDataSource = pep
+ .getIdentificationDataSource();
+ if (idDataSource.getClass().equals(IdentificationXtandemFile.class)) {
+ // cvparam specifique xtandem
+
+ this.writeCvParam("PROTICdbO:0000287", "" + pep.get_evalue(),
+ "xtandem peptide evalue");
+
+ this.writeCvParam("PROTICdbO:0000288", "" + pep.get_hypercorr(),
+ "xtandem peptide hyperscore");
+ }
+
+ this.writeCvParam("PROTICdbO:0000289",
+ "" + pep.getRetentionTimeInMinutes(),
+ "xtandem peptide retention time");
+ /*
+ * writer.writeStartElement("cvParam"); writer.writeAttribute("value",
+ * "" + pep.get_RT()); writer.writeAttribute("name",
+ * "xtandem peptide retention time"); writer.writeAttribute("cvLabel",
+ * "PROTICdbO"); writer.writeAttribute("accession",
+ * "PROTICdbO:0000289"); writer.writeEndElement();// cvParam
+ */
+
+ // Element cvParam4 = doc.createElement("cvParam");
+ // cvParam4.setAttribute("value", "" + pep.get_post());
+ // cvParam4.setAttribute("name", "xtandem peptide post");
+ // cvParam4.setAttribute("cvLabel", "PROTICdbO");
+ // cvParam4.setAttribute("accession", "PROTICdbO:00000XX");
+ // peptideHit.appendChild(cvParam4);
+ //
+ // Element cvParam5 = doc.createElement("cvParam");
+ // cvParam5.setAttribute("value", "" + pep.get_pre());
+ // cvParam5.setAttribute("name", "xtandem peptide pre");
+ // cvParam5.setAttribute("cvLabel", "PROTICdbO");
+ // cvParam5.setAttribute("accession", "PROTICdbO:00000XX");
+ // peptideHit.appendChild(cvParam5);
+
+ // sequences avec les modifs
+ writer.writeStartElement("pepSeq");
+ writer.writeStartElement("peptide");
+ writer.writeCharacters("" + pep.getSequence());
+ writer.writeEndElement();// peptide
+ if (pep.get_Modifs().size() != 0) {
+ this.writePtms(pep.get_Modifs());
+ }
+ writer.writeEndElement();// pepSeq
+
+ pepSampleScanCle2xmlId.put(pep.get_sample_scan_sequencepeptide_cle(),
+ pep_hit_id);
+ writer.writeEndElement();// peptideHit
+}
+
+private void writePtms(ArrayList<Modifs> mods) throws XMLStreamException {
+ writer.writeStartElement("ptms");
+ for (Modifs mod : mods) {
+ writer.writeStartElement("ptm");
+ writer.writeAttribute("diff_mono", "" + mod.get_modvalue());
+ writer.writeAttribute("position", "" + mod.get_posi());
+ writer.writeAttribute("aa", "" + mod.get_AA());
+ try {
+ this.writeOboModif(mod.getPsiMod());
+ } catch (MSMSException e) {
+ String message = "error writing OBO modif : " + e.getMessage();
+ logger.warn(message);
+ }
+ writer.writeEndElement();// ptm
+
+ }
+ writer.writeEndElement();// ptms
+}
+
+void ProticdbMl::writeOboModif(pappso::AaModificationP mod) {
+ _output_stream->writeStartElement("cvParam");
+ _output_stream->writeAttribute("name", mod->getName());
+ _output_stream->writeAttribute("cvLabel", "MOD");
+ _output_stream->writeAttribute("accession", mod->getAccession());
+ _output_stream->writeEndElement();// cvParam
+}
+
+private void writepeptideHits(Identification identification)
+throws MSMSException {
+ try {
+ writer.writeStartElement("peptideHits");
+ for (Group group : identification.getGrouping().getGroupList()) {
+ // generating peptide numbers
+ PeptideMassSet pepMassSet = new PeptideMassSet();
+
+ Hashtable<String, PeptideMass> pepSampleScanCle2peptideMass;
+ if (this.xtpExperimentType.equals("phospho")) {
+ pepSampleScanCle2peptideMass = pepMassSet
+ .numberingPhosphoGetSampleScanCle2peptideMass(group);
+
+ } else {
+ pepSampleScanCle2peptideMass = pepMassSet
+ .numberingGetSampleScanCle2peptideMass(group);
+ }
+
+ for (SubGroup sg : group.getSubGroupSet().getSubGroupList()) {
+ for (HashSampleScanSetProt hashProt : sg
+ .getHashSampleScanSetProtList()) {
+
+ ArrayList<Peptide> peptideList = null;
+ if (this.xtpExperimentType.equals("phospho")) {
+ peptideList = ((Match) hashProt.getGrpProtMatch())
+ .getValidPeptideInPhosphoIsland((HashSampleScanSetPhospho) hashProt);
+ } else {
+ peptideList = ((Match) hashProt.getGrpProtMatch())
+ .get_valid_peptides();
+ }
+
+ for (Peptide pep : peptideList) {
+
+ if (pepSampleScanCle2xmlId.containsKey(pep
+ .get_sample_scan_sequencepeptide_cle()) == false) {
+
+ writePeptideHit(
+ pep,
+ group,
+ pepSampleScanCle2peptideMass.get(pep
+ .get_sample_scan_sequencepeptide_cle()));
+
+ }
+ }
+ }
+ }
+ }
+ writer.writeEndElement();// "peptideHits");
+ } catch (Exception e) {
+ String message = "error writing peptideHits in ProticDbMl \n"
+ + e.getMessage();
+ logger.error(message);
+ throw new MSMSException(message);
+ }
+}
+
+private void writeMatchs(Identification identification)
+throws MSMSException {
+ try {
+ writer.writeStartElement("matchs");
+ for (Group group : identification.getGrouping().getGroupList()) {
+ for (SubGroup sg : group.getSubGroupSet().getSubGroupList()) {
+ writer.writeStartElement("match");
+
+ // echantillons
+ Hashtable<String, String> view = new Hashtable<String, String>();
+ for (HashSampleScanSetProt hashProt : sg
+ .getHashSampleScanSetProtList()) {
+ Match match = (Match) hashProt.getGrpProtMatch();
+ for (MsRun sample : match.get_sample_matchs()) {
+ if (!view.containsKey(sample.getSampleName())) {
+ writer.writeStartElement("matchSample");
+ writer.writeAttribute(
+ "sample_id",
+ ""
+ + sample_to_id.get(sample
+ .getSampleName()));
+ String name = sample.getSampleName();
+ name = name.replaceAll(".xml", "");
+ writer.writeAttribute("name", "" + name);
+ view.put(sample.getSampleName(), "ok");
+ writer.writeEndElement();// "matchSample");
+ }
+ }
+ }
+
+ // proteins
+ int k = 1;
+ for (HashSampleScanSetProt hashProt : sg
+ .getHashSampleScanSetProtList()) {
+ writer.writeStartElement("proteinHit");
+ Protein prot = (Protein) hashProt.getGrpProtMatch()
+ .getGrpProtein();
+ writer.writeAttribute("sequence_id",
+ "" + prot_to_id.get(prot.get_description()));
+ // proteinHit.setAttribute("score", "");
+ writer.writeAttribute("rank", "" + k);
+
+ // //cvparam
+ this.writeCvParam("PROTICdbO:0000284",
+ "" + prot.get_mw(),
+ "MW determination by Benuch");
+ /*
+ * writer.writeStartElement("cvParam");
+ * writer.writeAttribute("value", "" + prot.get_mw());
+ * writer.writeAttribute("name",
+ * "MW determination by Benuch");
+ * writer.writeAttribute("cvLabel", "PROTICdbO");
+ * writer.writeAttribute("accession",
+ * "PROTICdbO:0000284"); writer.writeEndElement();//
+ * cvParam
+ */
+ // evalue
+ MsRunSet samples = ((Match) hashProt.getGrpProtMatch())
+ .get_sample_matchs();
+ this.writeCvParam("PROTICdbO:0000291",
+ "" + prot.get_evalue(), "Xtandem log evalue");
+ /*
+ * writer.writeStartElement("cvParam");
+ * writer.writeAttribute("value", "" +
+ * prot.get_evalue());
+ *
+ * writer.writeAttribute("name", "Xtandem log evalue");
+ * writer.writeAttribute("cvLabel", "PROTICdbO");
+ * writer.writeAttribute("accession",
+ * "PROTICdbO:0000291"); writer.writeEndElement();//
+ * cvParam
+ */
+
+ // coverage
+ if (samples.size() == 1) {
+ this.writeCvParam(
+ "PROTICdbO:0000285",
+ ""
+ + ((Match) hashProt
+ .getGrpProtMatch())
+ .get_coverage(),
+ "protein coverage");
+ }
+
+ // [Term]
+ // id: PROTICdbO:0000335
+ // name: X!TandemPipeline PAI
+ this.writeCvParam("PROTICdbO:0000335",
+ "" + hashProt.getPAI(), "PAI");
+
+ // [Term]
+ // id: PROTICdbO:0000337
+ // name: protein group number
+ String pappsoGroupingNumber = Utils
+ .getPappsoGroupingNumber(group, sg, k);
+ if (pappsoGroupingNumber != null) {
+ this.writeCvParam("PROTICdbO:0000337",
+ pappsoGroupingNumber, "grouping number");
+ }
+
+ writer.writeEndElement();// "proteinHit");
+ k++;
+ }
+ for (HashSampleScanSetProt hashProt : sg
+ .getHashSampleScanSetProtList()) {
+ Protein prot2 = (Protein) hashProt.getGrpProtMatch()
+ .getGrpProtein();
+ // peptidesHitRef
+
+ ArrayList<Peptide> peptideList = null;
+ if (this.xtpExperimentType.equals("phospho")) {
+ peptideList = ((Match) hashProt.getGrpProtMatch())
+ .getValidPeptideInPhosphoIsland((HashSampleScanSetPhospho) hashProt);
+ } else {
+ peptideList = ((Match) hashProt.getGrpProtMatch())
+ .get_valid_peptides();
+ }
+
+ for (Peptide pep : peptideList) {
+ // for (Peptide pep : ((Match)
+ // hashProt.getGrpProtMatch())
+ // .get_peptide_order_inclus_redon()) {
+ String xmlId = pepSampleScanCle2xmlId.get(pep
+ .get_sample_scan_sequencepeptide_cle());
+ if (xmlId == null) {
+ String message = "the xml of peptide "
+ + pep.get_sample_scan_sequencepeptide_cle()
+ + " is not known in pepSampleScanCle2xmlId";
+ logger.debug(message);
+ throw new Exception(message);
+ }
+ writer.writeStartElement("peptideHitRef");
+ writer.writeAttribute("peptide_hit_id", "" + xmlId);
+
+ writer.writeStartElement("fromSeq");
+ writer.writeAttribute(
+ "seq_id",
+ ""
+ + prot_to_id.get(prot2
+ .get_description()));
+ writer.writeAttribute("start", "" + pep.get_start());
+ writer.writeAttribute("stop", "" + pep.get_stop());
+ String res1 = pep.get_pre();
+ res1 = res1.replaceAll("]", "-");
+ writer.writeAttribute("residue_before_nter", ""
+ + res1.charAt((res1.length() - 1)));
+ String res2 = pep.get_post();
+ res2 = res2.replaceAll("]", "-");
+ writer.writeAttribute("residue_after_cter", ""
+ + res2.charAt(0));
+
+ writer.writeEndElement();// fromSeq
+ writer.writeEndElement();// peptideHitRef
+ }
+
+ }
+ writer.writeEndElement();// "match");
+ }
+ }
+ writer.writeEndElement();// "matchs");
+ } catch (Exception e) {
+ String message = "error writing Matchs in ProticDbMl \n"
+ + e.getMessage();
+ logger.error(message);
+ throw new MSMSException(message);
+ }
+}
+
+void ProticdbMl::writeCvParam(QString acc, QString value, QString description) {
+ _output_stream->writeStartElement("cvParam");
+
+ _output_stream->writeAttribute("value",value);
+ if (description.isEmpty()) {
+ _output_stream->writeAttribute("name", "N.A.");
+ } else {
+ _output_stream->writeAttribute("name", description);
+ }
+ _output_stream->writeAttribute("cvLabel", "PROTICdbO");
+ _output_stream->writeAttribute("accession", acc);
+ _output_stream->writeEndElement();// cvParam
+
+}
diff --git a/src/output/proticdbml.h b/src/output/proticdbml.h
index e90b252f..8b226fd8 100644
--- a/src/output/proticdbml.h
+++ b/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;
diff --git a/src/utils/proteinstore.cpp b/src/utils/proteinstore.cpp
index 763693cc..6cf29b59 100644
--- a/src/utils/proteinstore.cpp
+++ b/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;
diff --git a/src/utils/types.h b/src/utils/types.h
index 432f5e15..9bf61756 100644
--- a/src/utils/types.h
+++ b/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
*
*/
diff --git a/src/utils/utils.cpp b/src/utils/utils.cpp
index b70d4398..9f4abcb4 100644
--- a/src/utils/utils.cpp
+++ b/src/utils/utils.cpp
@@ -1,8 +1,32 @@
#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;
+}
diff --git a/src/utils/utils.h b/src/utils/utils.h
index 3480d89d..47e451c0 100644
--- a/src/utils/utils.h
+++ b/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
--
GitLab