From a337dff2964199158ddec6cfb8511f5910c726f2 Mon Sep 17 00:00:00 2001
From: Olivier Langella <olivier.langella@u-psud.fr>
Date: Wed, 22 Nov 2017 22:27:49 +0100
Subject: [PATCH] WIP: new xpip parser
---
.../identificationdatasource.cpp | 3 +
.../identificationdatasource.h | 4 ++
src/input/xtpxpipsaxhandler.cpp | 68 +++++++++++++++++++
src/input/xtpxpipsaxhandler.h | 9 ++-
src/utils/identificationdatasourcestore.cpp | 40 +++++++++--
src/utils/identificationdatasourcestore.h | 1 +
6 files changed, 120 insertions(+), 5 deletions(-)
diff --git a/src/core/identification_sources/identificationdatasource.cpp b/src/core/identification_sources/identificationdatasource.cpp
index 3736af228..7899da8e9 100644
--- a/src/core/identification_sources/identificationdatasource.cpp
+++ b/src/core/identification_sources/identificationdatasource.cpp
@@ -75,6 +75,9 @@ void IdentificationDataSource::setMsRunSp (MsRunSp ms_run_sp) {
MsRunSp IdentificationDataSource::getMsRunSp () const {
return (_ms_run_sp);
}
+void IdentificationDataSource::setIdentificationEngine(IdentificationEngine engine) {
+ _engine = engine;
+}
IdentificationEngine IdentificationDataSource::getIdentificationEngine() const {
return _engine;
diff --git a/src/core/identification_sources/identificationdatasource.h b/src/core/identification_sources/identificationdatasource.h
index 30ca866e0..01a8d2202 100644
--- a/src/core/identification_sources/identificationdatasource.h
+++ b/src/core/identification_sources/identificationdatasource.h
@@ -70,6 +70,10 @@ public:
*/
virtual void parseTo(Project* p_project)=0;
+ /** \brief set identification engine
+ */
+ virtual void setIdentificationEngine(IdentificationEngine engine);
+
/** \brief identification engine
*/
virtual IdentificationEngine getIdentificationEngine() const;
diff --git a/src/input/xtpxpipsaxhandler.cpp b/src/input/xtpxpipsaxhandler.cpp
index b45ceb18d..90520333e 100644
--- a/src/input/xtpxpipsaxhandler.cpp
+++ b/src/input/xtpxpipsaxhandler.cpp
@@ -32,6 +32,7 @@
#include <pappsomspp/exception/exceptionnotfound.h>
#include <cmath>
+#include "../core/peptideevidence.h"
#include "../utils/peptidestore.h"
#include "../utils/proteinstore.h"
@@ -67,14 +68,20 @@ bool XtpXpipSaxHandler::startElement(const QString & namespaceURI, const QString
}
} else if (qName == "protein_match") {
is_ok = startElement_protein_match(attributes);
+ } else if (qName == "peptide_match") {
+ is_ok = startElement_peptide_match(attributes);
} else if (qName == "protein") {
is_ok = startElement_protein(attributes);
+ } else if (qName == "identification_source") {
+ is_ok = startElement_identification_source(attributes);
}
//<sample value="P6_08_10"/>
else if (qName == "sample") {
is_ok = startElement_msrun(attributes);
} else if (qName == "peptide") {
is_ok = startElement_peptide(attributes);
+ } else if (qName == "peptide_evidence") {
+ is_ok = startElement_peptide_evidence(attributes);
} else if (qName == "modification") {
is_ok = startElement_modification(attributes);
} else if (qName == "mod") {
@@ -187,6 +194,22 @@ bool XtpXpipSaxHandler::startElement_modification(QXmlAttributes attributes) {
return true;
}
+bool XtpXpipSaxHandler::startElement_identification_source(QXmlAttributes attributes) {
+ //<identification_source id="identa0" msrun_id="sampa0" path="/gorgone/pappso/formation/Janvier2014/TD/xml_tandem/20120906_balliau_extract_1_A01_urnb-1.xml" engine="1" version=""/>
+
+ qDebug() << "startElement_identification_source ";
+
+ IdentificationEngine engine = static_cast<IdentificationEngine>(attributes.value("engine").toUInt());
+ IdentificationDataSourceSp sp_ident_source = _p_project->getIdentificationDataSourceStore().getInstance(attributes.value("path").simplified(), engine);
+ sp_ident_source.get()->setXmlId(attributes.value("id").simplified());
+ _map_ident_sources.insert(std::pair<QString, IdentificationDataSourceSp>(sp_ident_source.get()->getXmlId(), sp_ident_source));
+
+ sp_ident_source->setMsRunSp(_map_msruns.at(attributes.value("msrun_id").simplified()));
+ sp_ident_source->setIdentificationEngineVersion(attributes.value("version"));
+ //_current_identification_group_p->addMsRunSp(ms_run);
+ qDebug() << "startElement_identification_source end" ;
+ return true;
+}
bool XtpXpipSaxHandler::startElement_msrun(QXmlAttributes attributes) {
//<msrun id="sampa0" name="20120906_balliau_extract_1_A01_urnb-1" format="0" path="20120906_balliau_extract_1_A01_urnb-1"/>
@@ -196,11 +219,56 @@ bool XtpXpipSaxHandler::startElement_msrun(QXmlAttributes attributes) {
ms_run.get()->setFilename(attributes.value("path").simplified());
ms_run.get()->setSampleName(attributes.value("name").simplified());
ms_run.get()->setXmlId(attributes.value("id").simplified());
+ _map_msruns.insert(std::pair<QString, MsRunSp>(ms_run.get()->getXmlId(), ms_run));
//_current_identification_group_p->addMsRunSp(ms_run);
qDebug() << "startElement_msrun end" ;
return true;
}
+bool XtpXpipSaxHandler::startElement_peptide_evidence(QXmlAttributes attributes) {
+
+ qDebug() << "startElement_peptide_evidence ";
+ /*
+ * <peptide_evidence id="peb76" source_id="identa0" peptide_id="pd7688" scan="11743" rt="2170" evalue="8.1e-10" exp_mass="1655.755648" charge="2" checked="true">
+ <param key="0" value="48.5"/>
+ </peptide_evidence>
+
+ */
+ IdentificationDataSourceSp sp_ident_source = _map_ident_sources.at(attributes.value("source_id").simplified());
+ unsigned int scan = attributes.value("scan").toUInt();
+ _p_peptide_evidence = new PeptideEvidence(sp_ident_source.get()->getMsRunSp().get(), scan);
+ _p_peptide_evidence->setIdentificationDataSource(sp_ident_source.get());
+
+ _p_peptide_evidence->setChecked(false);
+ if (attributes.value("checked").simplified().toLower() == "true") {
+ _p_peptide_evidence->setChecked(true);
+ }
+ _p_peptide_evidence->setCharge(attributes.value("charge").toUInt());
+ _p_peptide_evidence->setRetentionTime(attributes.value("rt").toDouble());
+ _p_peptide_evidence->setEvalue(attributes.value("evalue").toDouble());
+ _p_peptide_evidence->setExperimentalMass(attributes.value("exp_mass").toDouble());
+ _p_peptide_evidence->setPeptideXtpSp(_map_peptides.at( attributes.value("peptide_id").simplified()));
+
+ PeptideEvidenceSp sp_peptide_evidence = sp_ident_source.get()->getPeptideEvidenceStore().getInstance(_p_peptide_evidence);
+ _map_peptide_evidences.insert(std::pair<QString, PeptideEvidenceSp>(attributes.value("id").simplified(), sp_peptide_evidence));
+ qDebug() << "startElement_peptide_evidence end" ;
+ return true;
+}
+
+bool XtpXpipSaxHandler::startElement_peptide_match(QXmlAttributes attributes) {
+
+ qDebug() << "startElement_peptide_match ";
+ /*
+ * <protein_match acc="GRMZM2G138258_P01" checked="true">
+ <peptide_match source_id="identa0" peptide_id="pd3533" scan="1907" rt="646.406" evalue="0.00044" exp_mass="2190.816924" start="103" charge="3" checked="true">
+ <param key="0" value="23.6"/>
+ </peptide_match>
+ */
+ _p_protein_match = new ProteinMatch();
+ qDebug() << "startElement_peptide_match end" ;
+ return true;
+}
+
bool XtpXpipSaxHandler::startElement_protein_match(QXmlAttributes attributes) {
qDebug() << "startElement_protein_match ";
diff --git a/src/input/xtpxpipsaxhandler.h b/src/input/xtpxpipsaxhandler.h
index f8111f430..7aff217af 100644
--- a/src/input/xtpxpipsaxhandler.h
+++ b/src/input/xtpxpipsaxhandler.h
@@ -70,9 +70,12 @@ private:
bool startElement_description(QXmlAttributes attributes);
bool startElement_fasta_file(QXmlAttributes attributes);
bool startElement_protein_match(QXmlAttributes attributes);
+ bool startElement_peptide_match(QXmlAttributes attributes);
bool startElement_peptide(QXmlAttributes attributes);
+ bool startElement_peptide_evidence(QXmlAttributes attributes);
bool startElement_protein(QXmlAttributes attributes);
bool startElement_msrun(QXmlAttributes attributes);
+ bool startElement_identification_source(QXmlAttributes attributes);
bool startElement_modification(QXmlAttributes attributes);
bool startElement_mod(QXmlAttributes attributes);
//bool endElement_identification();
@@ -90,12 +93,16 @@ private:
Project * _p_project;
ProteinMatch * _p_protein_match;
- PeptideMatch * _p_peptide_match;
+ PeptideMatch _current_peptide_match;
+ PeptideEvidence * _p_peptide_evidence;
IdentificationGroup * _current_identification_group_p;
std::map<QString, pappso::AaModificationP> _map_modifs;
std::map<QString, FastaFileSp> _map_fasta_files;
std::map<QString, ProteinXtpSp> _map_proteins;
+ std::map<QString, MsRunSp> _map_msruns;
+ std::map<QString, IdentificationDataSourceSp> _map_ident_sources;
+ std::map<QString, PeptideEvidenceSp> _map_peptide_evidences;
ProteinXtp _current_protein;
PeptideXtpSp _current_peptide_sp;
std::map<QString, PeptideXtpSp> _map_peptides;
diff --git a/src/utils/identificationdatasourcestore.cpp b/src/utils/identificationdatasourcestore.cpp
index c01d92ac1..46059e90c 100644
--- a/src/utils/identificationdatasourcestore.cpp
+++ b/src/utils/identificationdatasourcestore.cpp
@@ -44,6 +44,39 @@ IdentificationDataSourceStore::~IdentificationDataSourceStore()
}
+IdentificationDataSourceSp IdentificationDataSourceStore::getInstance(const QString & location, IdentificationEngine engine) {
+ qDebug() << "IdentificationDataSourceStore::getInstance begin " << location;
+ qDebug() << " " << _map_identification_data_sources.size();
+ std::map< QString, IdentificationDataSourceSp >::iterator it = _map_identification_data_sources.find(location);
+ if (it != _map_identification_data_sources.end()) {
+ return it->second;
+ }
+ else {
+ QFileInfo location_file(location);
+ QString ext = location_file.suffix();
+ //QString sample_name = location_file.baseName();
+ IdentificationDataSourceSp p_identfile = nullptr;
+ if (ext.toLower() == "xml") {
+ //X!Tandem result file
+ p_identfile = std::make_shared<IdentificationXtandemFile>(location_file);
+ } else if (ext.toLower() == "pep") {
+ //pep xml file
+ p_identfile = std::make_shared<IdentificationPwizFile>(location_file);
+ } else {
+ p_identfile = std::make_shared<IdentificationPwizFile>(location_file);
+ }
+ if (p_identfile == nullptr) {
+ throw pappso::PappsoException(QObject::tr("Identification resource %1 not recognized (null pointer)").arg(location));
+ }
+ p_identfile.get()->setXmlId(QString("ident%1").arg(pappso::Utils::getLexicalOrderedString(_map_identification_data_sources.size())));
+ p_identfile.get()->setIdentificationEngine(engine);
+ _map_identification_data_sources.insert(std::pair< QString, IdentificationDataSourceSp >(location, p_identfile));
+ _map_identification_data_sources.insert(std::pair< QString, IdentificationDataSourceSp >(location_file.absoluteFilePath(), p_identfile));
+ return p_identfile;
+ }
+ throw pappso::PappsoException(QObject::tr("Identification resource %1 not recognized").arg(location));
+}
+
IdentificationDataSourceSp IdentificationDataSourceStore::getInstance(const QString & location) {
qDebug() << "IdentificationDataSourceStore::getInstance begin " << location;
qDebug() << " " << _map_identification_data_sources.size();
@@ -59,18 +92,17 @@ IdentificationDataSourceSp IdentificationDataSourceStore::getInstance(const QStr
if (ext.toLower() == "xml") {
//X!Tandem result file
p_identfile = std::make_shared<IdentificationXtandemFile>(location_file);
- } else
- if (ext.toLower() == "pep") {
+ } else if (ext.toLower() == "pep") {
//pep xml file
p_identfile = std::make_shared<IdentificationPwizFile>(location_file);
- }else {
+ } else {
p_identfile = std::make_shared<IdentificationPwizFile>(location_file);
}
if (p_identfile == nullptr) {
throw pappso::PappsoException(QObject::tr("Identification resource %1 not recognized (null pointer)").arg(location));
}
p_identfile.get()->setXmlId(QString("ident%1").arg(pappso::Utils::getLexicalOrderedString(_map_identification_data_sources.size())));
-
+
_map_identification_data_sources.insert(std::pair< QString, IdentificationDataSourceSp >(location, p_identfile));
_map_identification_data_sources.insert(std::pair< QString, IdentificationDataSourceSp >(location_file.absoluteFilePath(), p_identfile));
return p_identfile;
diff --git a/src/utils/identificationdatasourcestore.h b/src/utils/identificationdatasourcestore.h
index a2e204f34..a84b06a83 100644
--- a/src/utils/identificationdatasourcestore.h
+++ b/src/utils/identificationdatasourcestore.h
@@ -42,6 +42,7 @@ public:
IdentificationDataSourceStore();
~IdentificationDataSourceStore();
IdentificationDataSourceSp getInstance(const QString & location);
+ IdentificationDataSourceSp getInstance(const QString & location, IdentificationEngine engine);
std::vector<IdentificationDataSourceSp> getIdentificationDataSourceList() const;
private :
--
GitLab