From a71b121197035a5d1dda43453761b3dc9d97098c Mon Sep 17 00:00:00 2001
From: Olivier Langella <Olivier.Langella@moulon.inra.fr>
Date: Wed, 17 May 2017 21:31:08 +0200
Subject: [PATCH] WIP: proticdbml output
---
src/core/identificationgroup.cpp | 3 +
src/core/identificationgroup.h | 1 +
src/output/proticdbml.cpp | 361 ++++++++++++++++---------------
src/output/proticdbml.h | 1 +
4 files changed, 196 insertions(+), 170 deletions(-)
diff --git a/src/core/identificationgroup.cpp b/src/core/identificationgroup.cpp
index 78e09767a..1dbc4b558 100644
--- a/src/core/identificationgroup.cpp
+++ b/src/core/identificationgroup.cpp
@@ -166,6 +166,9 @@ void IdentificationGroup::addMsRunSp(MsRunSp ms_run_sp) {
const std::vector<MsRunSp> & IdentificationGroup::getMsRunSpList() const {
return _ms_run_list;
}
+const std::vector<IdentificationDataSource *> & IdentificationGroup::getIdentificationDataSourceList() const {
+ return _id_source_list;
+}
std::vector<ProteinMatch *> & IdentificationGroup::getProteinMatchList() {
return _protein_match_list;
}
diff --git a/src/core/identificationgroup.h b/src/core/identificationgroup.h
index 0d5de141e..1ed50884c 100644
--- a/src/core/identificationgroup.h
+++ b/src/core/identificationgroup.h
@@ -93,6 +93,7 @@ public:
void startGrouping (const GroupingType & grouping_type);
const std::vector<MsRunSp> & getMsRunSpList() const;
+ const std::vector<IdentificationDataSource *> & getIdentificationDataSourceList() const;
/** @brief get tab name for qtabwidget
* */
diff --git a/src/output/proticdbml.cpp b/src/output/proticdbml.cpp
index 06ba2c163..371b400b5 100644
--- a/src/output/proticdbml.cpp
+++ b/src/output/proticdbml.cpp
@@ -33,6 +33,7 @@
#include <pappsomspp/grouping/grpprotein.h>
#include <pappsomspp/amino_acid/Aa.h>
#include "../core/msrun.h"
+#include "../config.h"
ProticdbMl::ProticdbMl(const QString & out_filename)
{
@@ -149,7 +150,7 @@ void ProticdbMl::writeSequence(ProteinMatch * p_protein_match) {
}
void ProticdbMl::writeProject() {
-qDebug() << "ProticdbMl::writeProject begin";
+ qDebug() << "ProticdbMl::writeProject begin";
_output_stream->writeStartElement("project");
_output_stream->writeAttribute("name", "");
_output_stream->writeAttribute("id", "p1");
@@ -178,101 +179,66 @@ qDebug() << "ProticdbMl::writeProject begin";
_output_stream->writeEndDocument();
-qDebug() << "ProticdbMl::writeProject end duration = " << duracel << "ms";
+ 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 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();
+ // Ajout des base de données
+ _output_stream->writeStartElement("customDb");
+ _output_stream->writeAttribute("name", "");
+ _output_stream->writeAttribute("id", "customdb0");
+ _output_stream->writeEndElement();
}
void ProticdbMl::writeSamples() {
- _output_stream->writeStartElement("samples");
- for (std::vector<MsRunSp> msrun_sp : _sp_project.get()->getMsRunStore().getMsRunList()) {
- _output_stream->writeStartElement("sample");
- QString id_samp = "samp" + pappso::Utils::getLexicalOrderedString(sample_to_id.size());
- QString id_msrun = msrun_sp.get()->getXmlId();
- QString name = msrun_sp.get()->getSampleName();
- // balise sample
- _output_stream->writeAttribute("name", name);
- _output_stream->writeAttribute("id", id_samp);
- _output_stream->writeEmptyElement("description");
- _sample_to_id.insert(std::pair<QString, QString>( name, id_samp));
-
- // Element spectrumList = doc.createElement("spectrumList");
- // msRun.appendChild(spectrumList);
-
- //msrun_to_id.put(samp.getSampleName(), id_msrun);
- _output_stream->writeEndElement();// sample
-
- }
- _output_stream->writeEndElement();// samples
+ _output_stream->writeStartElement("samples");
+ for (std::vector<MsRunSp> msrun_sp : _sp_project.get()->getMsRunStore().getMsRunList()) {
+ _output_stream->writeStartElement("sample");
+ QString id_samp = "samp" + pappso::Utils::getLexicalOrderedString(sample_to_id.size());
+ QString id_msrun = msrun_sp.get()->getXmlId();
+ QString name = msrun_sp.get()->getSampleName();
+ // balise sample
+ _output_stream->writeAttribute("name", name);
+ _output_stream->writeAttribute("id", id_samp);
+ _output_stream->writeEmptyElement("description");
+ _sample_to_id.insert(std::pair<QString, QString>( name, id_samp));
+
+ // Element spectrumList = doc.createElement("spectrumList");
+ // msRun.appendChild(spectrumList);
+
+ //msrun_to_id.put(samp.getSampleName(), id_msrun);
+ _output_stream->writeEndElement();// sample
+
+ }
+ _output_stream->writeEndElement();// samples
}
void ProticdbMl::writeMsRuns() {
-
- _output_stream->writeStartElement("msRuns");
- for (std::vector<MsRunSp> msrun_sp : _sp_project.get()->getMsRunStore().getMsRunList()) {
-
- QString name = msrun_sp.get()->getSampleName();
-
- QString id_samp = _sample_to_id.at(name);
-
- // Element spectrumList = doc.createElement("spectrumList");
- // msRun.appendChild(spectrumList);
-
- QString id_msrun = msrun_sp.get()->getXmlId();
-
- // balise MsRun
- _output_stream->writeStartElement("msRun");
- _output_stream->writeAttribute("sample_id", id_samp);
- _output_stream->writeAttribute("id", id_msrun);
-
- _output_stream->writeStartElement("description");
- _output_stream->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
-
- _output_stream->writeStartElement("sourceFile");
- _output_stream->writeStartElement("nameOfFile");
- _output_stream->writeCharacters(name + ".RAW");
- _output_stream->writeEndElement();// nameOfFile
- _output_stream->writeEndElement();// sourceFile
-
- _output_stream->writeStartElement("sampleName");
- _output_stream->writeCharacters(name);
- _output_stream->writeEndElement();// sampleName
- _output_stream->writeEndElement();// admin
- _output_stream->writeEndElement();// description
-
- _output_stream->writeEndElement();// msRun
-
- }
- _output_stream->writeEndElement();// msRuns
-}
+ _output_stream->writeStartElement("msRuns");
+ for (std::vector<MsRunSp> msrun_sp : _sp_project.get()->getMsRunStore().getMsRunList()) {
-void ProticdbMl::writeIdentificationRun(IdentificationGroup * p_identification) {
- _output_stream->writeAttribute("ident_method_id", "m1");
- _output_stream->writeAttribute("customdb_id", "customdb0");
+ QString name = msrun_sp.get()->getSampleName();
+
+ QString id_samp = _sample_to_id.at(name);
+
+ // Element spectrumList = doc.createElement("spectrumList");
+ // msRun.appendChild(spectrumList);
+
+ QString id_msrun = msrun_sp.get()->getXmlId();
+
+ // balise MsRun
+ _output_stream->writeStartElement("msRun");
+ _output_stream->writeAttribute("sample_id", id_samp);
+ _output_stream->writeAttribute("id", id_msrun);
_output_stream->writeStartElement("description");
_output_stream->writeStartElement("admin");
@@ -285,108 +251,163 @@ void ProticdbMl::writeIdentificationRun(IdentificationGroup * p_identification)
// writer.writeCharacters("Valot Benoit");
// writer.writeEndElement();// name
// writer.writeEndElement();// contact
- /*
- this.xpipFile = XtandemPipelineSession.getInstance()
- .getCurrentXpipFile();
_output_stream->writeStartElement("sourceFile");
_output_stream->writeStartElement("nameOfFile");
- // TODO
- // get the loaded xpip file name
- _output_stream->writeCharacters(this.xpipFile.getName());
+ _output_stream->writeCharacters(name + ".RAW");
_output_stream->writeEndElement();// nameOfFile
- _output_stream->writeStartElement("pathToFile");
- _output_stream->writeCharacters(this.xpipFile.getAbsolutePath());
- _output_stream->writeEndElement();// pathToFile
- _output_stream->writeStartElement("fileType");
- _output_stream->writeCharacters("XPIP file");
- _output_stream->writeEndElement();// fileType
_output_stream->writeEndElement();// sourceFile
-*/
+
+ _output_stream->writeStartElement("sampleName");
+ _output_stream->writeCharacters(name);
+ _output_stream->writeEndElement();// sampleName
_output_stream->writeEndElement();// admin
+ _output_stream->writeEndElement();// description
- // 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
- }
+ _output_stream->writeEndElement();// msRun
- // 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
-
- 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, "");
- }
+ }
+ _output_stream->writeEndElement();// msRuns
+}
+
+void ProticdbMl::writeIdentificationDataSource(IdentificationDataSource * p_identification_data_source) {
+
+
+ _output_stream->writeStartElement("dataProcessing");
+ _output_stream->writeStartElement("software");
+ _output_stream->writeStartElement("name");
+ if (p_identification_data_source->getIdentificationEngine() == IdentificationEngine::XTandem) {
+ _output_stream->writeAttribute("acc", "PROTICdbO:0000283");
+ }
+ _output_stream->writeCharacters(p_identification_data_source->getIdentificationEngineName());
+
+ _output_stream->writeEndElement();// name
+ _output_stream->writeStartElement("version");
+ _output_stream->writeCharacters(p_identification_data_source->getIdentificationEngineVersion());
+ _output_stream->writeEndElement();// version
+ _output_stream->writeEndElement();// software
+ _output_stream->writeStartElement("processingMethod");
+ // mzXml source file name
+ writeCvParam("PROTICdbO:0000343",p_identification_data_source->getMsRunSp().get()->getFilename(),
+ "MS/MS data source file name");
+ if (p_identification_data_source->getIdentificationEngine() == IdentificationEngine::XTandem) {
+ // model file name
+ writeCvParam("PROTICdbO:0000342", p_identification_data_source->getIdentificationEngineParam(IdentificationEngineParam::tandem_param).toString(),
+ "X!tandem xml model file name");
+ // xtandem result file name
+ writeCvParam("PROTICdbO:0000341", p_identification_data_source->getResourceName(),
+ "X!tandem xml result file name");
+ }
+ _output_stream->writeEndElement();// processingMethod
+ _output_stream->writeEndElement();// dataProcessing
+ /*
+ _output_stream->writeStartElement("dataProcessing");
+ _output_stream->writeStartElement("software");
+ _output_stream->writeStartElement("name");
+ _output_stream->writeAttribute("acc", "PROTICdbO:0000283");
+ _output_stream->writeCharacters("X!Tandem");
+ _output_stream->writeEndElement();// name
+ _output_stream->writeStartElement("version");
+ _output_stream->writeCharacters(XtandemPipelineSession.getInstance()
+ .getConfig().getXtandemVersion());
+ _output_stream->writeEndElement();// version
+ _output_stream->writeEndElement();// software
+ _output_stream->writeStartElement("processingMethod");
+ _output_stream->writeEndElement();// processingMethod
+ _output_stream->writeEndElement();// dataProcessing
+ */
+}
+
+void ProticdbMl::writeIdentificationRun(IdentificationGroup * p_identification) {
+ _output_stream->writeAttribute("ident_method_id", "m1");
+ _output_stream->writeAttribute("customdb_id", "customdb0");
+
+ _output_stream->writeStartElement("description");
+ _output_stream->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();
+
+ _output_stream->writeStartElement("sourceFile");
+ _output_stream->writeStartElement("nameOfFile");
+ // TODO
+ // get the loaded xpip file name
+ _output_stream->writeCharacters(this.xpipFile.getName());
+ _output_stream->writeEndElement();// nameOfFile
+ _output_stream->writeStartElement("pathToFile");
+ _output_stream->writeCharacters(this.xpipFile.getAbsolutePath());
+ _output_stream->writeEndElement();// pathToFile
+ _output_stream->writeStartElement("fileType");
+ _output_stream->writeCharacters("XPIP file");
+ _output_stream->writeEndElement();// fileType
+ _output_stream->writeEndElement();// sourceFile
+ */
+ _output_stream->writeEndElement();// admin
+
+ // if we can retrieve original informations in xml xtandem results
+ //
+ for (IdentificationDataSource * p_identification_data_source : p_identification->getIdentificationDataSourceList()) {
+ writeIdentificationDataSource(p_identification_data_source);
+ }
+
+// id: PROTICdbO:0000316
+// name: X!TandemPipeline
+
+ _output_stream->writeStartElement("dataProcessing");
+ _output_stream->writeStartElement("software");
+ _output_stream->writeStartElement("name");
+ _output_stream->writeAttribute("acc", "PROTICdbO:0000316");
+ _output_stream->writeCharacters("X!TandemPipeline");
+ _output_stream->writeEndElement();// name
+ _output_stream->writeStartElement("version");
+ _output_stream->writeCharacters(XTPCPP_VERSION);
+ _output_stream->writeEndElement();// version
+ _output_stream->writeEndElement();// software
+ _output_stream->writeStartElement("processingMethod");
+// cvParams
+
+ writeCvParam("PROTICdbO:0000323", QString("%1").arg(_sp_project.get()->getAutomaticFilterParameters().getFilterProteinEvalue()),
+ "X!TandemPipeline filter on protein evalue (log)");
+
+// indi
+// combine
+// phospho
+ if (_sp_project.get()->isCombineMode()) {
+ writeCvParam("PROTICdbO:0000320", "combine", "");
+ }
+ else {
+ writeCvParam("PROTICdbO:0000319", "indiv", "");
+ }
+ //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");
+ writeCvParam("PROTICdbO:0000325", QString("%1").arg(_sp_project.get()->getAutomaticFilterParameters().getFilterMinimumPeptidePerMatch()),
+ "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");
+ writeCvParam("PROTICdbO:0000324", QString("%1").arg(_sp_project.get()->getAutomaticFilterParameters().getFilterPeptideEvalue()),
+ "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");
+// TODO write database filter
+// this.writeCvParam("PROTICdbO:0000324", ""
+// +
+// XtandemPipelineSession.getInstance().getConfig().get_database_filter(),
+// "X!TandemPipeline filter on peptide evalue");
- writer.writeEndElement();// processingMethod
+ _output_stream->writeEndElement();// processingMethod
- writer.writeEndElement();// dataProcessing
+ _output_stream->writeEndElement();// dataProcessing
- writer.writeEndElement();// description
+ _output_stream->writeEndElement();// description
}
private void writePeptideHit(Peptide pep, Group group,
diff --git a/src/output/proticdbml.h b/src/output/proticdbml.h
index b3364df62..d1c58b429 100644
--- a/src/output/proticdbml.h
+++ b/src/output/proticdbml.h
@@ -56,6 +56,7 @@ private :
void writeSamples();
void writeMsRuns();
void writeIdentificationRun(IdentificationGroup * p_identification);
+ void writeIdentificationDataSource(IdentificationDataSource * p_identification_data_source);
private :
QFile * _output_file;
--
GitLab