From 038d50da7cfa37b68d4789f0362a75b8fbb17f46 Mon Sep 17 00:00:00 2001
From: Olivier Langella <Olivier.Langella@moulon.inra.fr>
Date: Mon, 25 Sep 2017 11:34:53 +0200
Subject: [PATCH] new identification engine statistics
---
.../identificationdatasource.cpp | 13 ++++++
.../identificationdatasource.h | 12 ++++++
.../tandem_run_dialog/tandem_run_dialog.ui | 19 +--------
src/gui/tandem_run_dialog/tandemrundialog.cpp | 4 ++
src/gui/tandem_run_dialog/tandemrundialog.h | 1 +
src/input/xtandemsaxhandler.cpp | 41 +++++++++++++------
src/output/ods/samplesheet.cpp | 32 +++++++++++++++
src/utils/types.h | 13 ++++++
8 files changed, 106 insertions(+), 29 deletions(-)
diff --git a/src/core/identification_sources/identificationdatasource.cpp b/src/core/identification_sources/identificationdatasource.cpp
index dcb377db8..c49f027c5 100644
--- a/src/core/identification_sources/identificationdatasource.cpp
+++ b/src/core/identification_sources/identificationdatasource.cpp
@@ -97,6 +97,19 @@ const QVariant IdentificationDataSource::getIdentificationEngineParam(Identifica
return QVariant();
}
}
+
+void IdentificationDataSource::setIdentificationEngineStatistics(IdentificationEngineStatistics param, const QVariant& value) {
+ _param_stats.insert(std::pair<IdentificationEngineStatistics, QVariant>(param, value));
+}
+const QVariant IdentificationDataSource::getIdentificationEngineStatistics(IdentificationEngineStatistics param) const {
+ try {
+ return _param_stats.at(param);
+ }
+ catch (std::out_of_range) {
+ return QVariant();
+ }
+}
+
pappso::SpectrumSp IdentificationDataSource::getSpectrumSp(unsigned int scan_number) const {
pappso::SpectrumSp spectrum_sp = SpectrumStore::getSpectrumSpFromMsRunSp(_ms_run_sp, scan_number);
return spectrum_sp;
diff --git a/src/core/identification_sources/identificationdatasource.h b/src/core/identification_sources/identificationdatasource.h
index daf0ab5c1..f99f8720b 100644
--- a/src/core/identification_sources/identificationdatasource.h
+++ b/src/core/identification_sources/identificationdatasource.h
@@ -85,7 +85,18 @@ public:
/** \brief get specific identification engine parameter value
*/
virtual const QVariant getIdentificationEngineParam(IdentificationEngineParam param) const;
+
+
+ /** \brief set identification engine statistics
+ * any statistics on this identification run that can be told by the identification engine
+ */
+ virtual void setIdentificationEngineStatistics(IdentificationEngineStatistics param, const QVariant& value);
+
+ /** \brief get specific identification engine statistics value
+ */
+ virtual const QVariant getIdentificationEngineStatistics(IdentificationEngineStatistics param) const;
+
/** \brief add Fastafile used by the identification engine
*/
void addFastaFile (FastaFileSp file);
@@ -100,6 +111,7 @@ private :
QString _version;
MsRunSp _ms_run_sp = nullptr;
std::map<IdentificationEngineParam, QVariant> _params;
+ std::map<IdentificationEngineStatistics, QVariant> _param_stats;
std::vector<FastaFileSp> _fastafile_list;
};
diff --git a/src/gui/tandem_run_dialog/tandem_run_dialog.ui b/src/gui/tandem_run_dialog/tandem_run_dialog.ui
index 6080f6325..79fa385b9 100644
--- a/src/gui/tandem_run_dialog/tandem_run_dialog.ui
+++ b/src/gui/tandem_run_dialog/tandem_run_dialog.ui
@@ -253,22 +253,6 @@
</widget>
<resources/>
<connections>
- <connection>
- <sender>add_files_button</sender>
- <signal>clicked()</signal>
- <receiver>TandemRunDialog</receiver>
- <slot>chooseFiles()</slot>
- <hints>
- <hint type="sourcelabel">
- <x>550</x>
- <y>310</y>
- </hint>
- <hint type="destinationlabel">
- <x>550</x>
- <y>264</y>
- </hint>
- </hints>
- </connection>
<connection>
<sender>pushButton_2</sender>
<signal>clicked()</signal>
@@ -305,7 +289,7 @@
<sender>clear_list_button</sender>
<signal>clicked()</signal>
<receiver>TandemRunDialog</receiver>
- <slot>clearFileList()</slot>
+ <slot>clearFastaFiles()</slot>
<hints>
<hint type="sourcelabel">
<x>464</x>
@@ -402,6 +386,7 @@
<slot>reject()</slot>
<slot>accept()</slot>
<slot>selectFastaFiles()</slot>
+ <slot>clearFastaFiles()</slot>
<slot>selectMzFiles()</slot>
<slot>selectOutputDirectory()</slot>
<slot>setPresetName(QString)</slot>
diff --git a/src/gui/tandem_run_dialog/tandemrundialog.cpp b/src/gui/tandem_run_dialog/tandemrundialog.cpp
index 50302ebbd..2a6ff3ebc 100644
--- a/src/gui/tandem_run_dialog/tandemrundialog.cpp
+++ b/src/gui/tandem_run_dialog/tandemrundialog.cpp
@@ -159,6 +159,10 @@ void TandemRunDialog::selectOutputDirectory() {
}
}
+void TandemRunDialog::clearFastaFiles() {
+ _p_fasta_file_list->removeRows( 0, _p_fasta_file_list->rowCount() );
+}
+
void TandemRunDialog::selectFastaFiles() {
try {
QSettings settings;
diff --git a/src/gui/tandem_run_dialog/tandemrundialog.h b/src/gui/tandem_run_dialog/tandemrundialog.h
index f2601144d..83944614e 100644
--- a/src/gui/tandem_run_dialog/tandemrundialog.h
+++ b/src/gui/tandem_run_dialog/tandemrundialog.h
@@ -51,6 +51,7 @@ public:
void reset();
public slots:
void selectFastaFiles();
+ void clearFastaFiles();
void selectMzFiles();
void selectOutputDirectory();
void selectPresetDirectory();
diff --git a/src/input/xtandemsaxhandler.cpp b/src/input/xtandemsaxhandler.cpp
index 2226d76e4..165619288 100644
--- a/src/input/xtandemsaxhandler.cpp
+++ b/src/input/xtandemsaxhandler.cpp
@@ -295,7 +295,7 @@ bool XtandemSaxHandler::startElement_domain(QXmlAttributes attributes) {
_p_peptide_match->setExperimentalMass(exp_mass);
_p_peptide_match->setStart(attributes.value("start").simplified().toUInt()-1);
_p_peptide_match->setCharge(_charge);
-
+
_p_peptide_match->setParam(PeptideMatchParam::tandem_hyperscore, QVariant( attributes.value("hyperscore").toDouble()));
_p_peptide_match->setIdentificationDataSource( _p_identification_data_source);
@@ -336,11 +336,11 @@ bool XtandemSaxHandler::endElement_note() {
bool is_ok = true;
if (_is_protein_description) {
//_p_protein_match->getProteinXtpSp().get()->setDescription(_current_text.section(" ",1));
- _p_protein_match->getProteinXtpSp().get()->setCompleteDescription(_current_text);
- if (!_p_protein_match->getProteinXtpSp().get()->getAccession().endsWith(":reversed") && _p_protein_match->getProteinXtpSp().get()->getDescription().endsWith(":reversed")) {
- //to fit most cases, just check that the :reversed chars added by X!Tandem are not in the description. if so, then add it too in the accession
- _p_protein_match->getProteinXtpSp().get()->setAccession(QString("%1%2").arg(_p_protein_match->getProteinXtpSp().get()->getAccession()).arg(":reversed"));
- }
+ _p_protein_match->getProteinXtpSp().get()->setCompleteDescription(_current_text);
+ if (!_p_protein_match->getProteinXtpSp().get()->getAccession().endsWith(":reversed") && _p_protein_match->getProteinXtpSp().get()->getDescription().endsWith(":reversed")) {
+ //to fit most cases, just check that the :reversed chars added by X!Tandem are not in the description. if so, then add it too in the accession
+ _p_protein_match->getProteinXtpSp().get()->setAccession(QString("%1%2").arg(_p_protein_match->getProteinXtpSp().get()->getAccession()).arg(":reversed"));
+ }
}
else {
@@ -470,13 +470,30 @@ bool XtandemSaxHandler::endElement_note() {
<note label="modelling, estimated false positives">18</note>
<note label="modelling, reversed sequence false positives">20</note>
<note label="modelling, spectrum noise suppression ratio">0.00</note>
- <note label="modelling, total peptides used">96618641</note>
- <note label="modelling, total proteins used">273656</note>
- <note label="modelling, total spectra assigned">7464</note>
- <note label="modelling, total spectra used">12199</note>
- <note label="modelling, total unique assigned">6260</note>
- <note label="process, start time">2013:12:20:16:47:19</note>
*/
+ //<note label="modelling, total peptides used">96618641</note>
+ if (_current_note_label == "modelling, total peptides used") {
+ _p_identification_data_source->setIdentificationEngineStatistics(IdentificationEngineStatistics::total_peptide_used, _current_text.toUInt());
+ }
+
+ //<note label="modelling, total proteins used">273656</note>
+ if (_current_note_label == "modelling, total proteins used") {
+ _p_identification_data_source->setIdentificationEngineStatistics(IdentificationEngineStatistics::total_proteins_used, _current_text.toUInt());
+ }
+ //<note label="modelling, total spectra assigned">7464</note>
+ if (_current_note_label == "modelling, total spectra assigned") {
+ _p_identification_data_source->setIdentificationEngineStatistics(IdentificationEngineStatistics::total_spectra_assigned, _current_text.toUInt());
+ }
+ //<note label="modelling, total spectra used">12199</note>
+ if (_current_note_label == "modelling, total spectra used") {
+ _p_identification_data_source->setIdentificationEngineStatistics(IdentificationEngineStatistics::total_spectra_used, _current_text.toUInt());
+ }
+ //<note label="modelling, total unique assigned">6260</note>
+ if (_current_note_label == "modelling, total unique assigned") {
+ _p_identification_data_source->setIdentificationEngineStatistics(IdentificationEngineStatistics::total_unique_assigned, _current_text.toUInt());
+ }
+ //<note label="process, start time">2013:12:20:16:47:19</note>
+
//<note label="process, version">X! Tandem Sledgehammer (2013.09.01.1)</note>
if (_current_note_label == "process, version") {
QRegExp rx("\\((.*)\\)");
diff --git a/src/output/ods/samplesheet.cpp b/src/output/ods/samplesheet.cpp
index dbd76c165..2a406f717 100644
--- a/src/output/ods/samplesheet.cpp
+++ b/src/output/ods/samplesheet.cpp
@@ -60,6 +60,29 @@ void SampleSheet::writeHeaders() {
_p_writer->writeCell("Identification fasta files");
_p_writer->writeCell("X!Tandem parameters");
+ // total_spectra_assigned=1, ///< total_spectra_assigned in one identification file (one sample)
+ _p_writer->setCellAnnotation("total spectra assigned in one identification file (one mzdata sample) given by the identification engine");
+ _p_writer->writeCell("Total spectra assigned");
+ //
+ //total_spectra_used=2,///< total_spectra_used in one identification file (one sample)
+ _p_writer->setCellAnnotation("total spectra used in one identification file (one mzdata sample) given by the identification engine");
+ _p_writer->writeCell("Total spectra used");
+
+
+ _p_writer->setCellAnnotation("Percentage of spectra assignment (spectra assigned / spectra used");
+ _p_writer->writeCell("Assignment percentage");
+
+ //total_unique_assigned=5,///< total number unique peptide sequence assigned
+ _p_writer->setCellAnnotation("total unique peptide sequence assigned in one identification file (one mzdata sample) given by the identification engine");
+ _p_writer->writeCell("Total unique assigned");
+
+
+
+ /*
+ total_peptide_used=3,///< total number of peptides generated and used in identification
+ total_proteins_used=4,///< total number of proteins generated and used in identification
+
+ */
}
@@ -79,5 +102,14 @@ void SampleSheet::writeIdentificationDataSource(IdentificationDataSource * p_ide
_p_writer->writeCell(fasta_files.join(" "));
_p_writer->writeCell(p_ident_data_source->getIdentificationEngineParam(IdentificationEngineParam::tandem_param).toString());
+ _p_writer->writeCell(p_ident_data_source->getIdentificationEngineStatistics(IdentificationEngineStatistics::total_spectra_assigned).toString());
+
+ _p_writer->writeCell(p_ident_data_source->getIdentificationEngineStatistics(IdentificationEngineStatistics::total_spectra_used).toString());
+
+ if (!p_ident_data_source->getIdentificationEngineStatistics(IdentificationEngineStatistics::total_spectra_assigned).isNull() && !p_ident_data_source->getIdentificationEngineStatistics(IdentificationEngineStatistics::total_spectra_used).isNull()) {
+ _p_writer->writeCellPercentage(p_ident_data_source->getIdentificationEngineStatistics(IdentificationEngineStatistics::total_spectra_assigned).toDouble() / p_ident_data_source->getIdentificationEngineStatistics(IdentificationEngineStatistics::total_spectra_used).toDouble());
+ }
+
+ _p_writer->writeCell(p_ident_data_source->getIdentificationEngineStatistics(IdentificationEngineStatistics::total_unique_assigned).toString());
}
diff --git a/src/utils/types.h b/src/utils/types.h
index b18293cd2..a264e3aaf 100644
--- a/src/utils/types.h
+++ b/src/utils/types.h
@@ -64,6 +64,19 @@ enum class IdentificationEngineParam {
tandem_param ///< X!Tandem xml parameters file
};
+
+/** \def IdentificationEngineStatistics identification engine statistics
+ *
+ */
+enum class IdentificationEngineStatistics: std::int8_t {
+ total_spectra_assigned=1, ///< total_spectra_assigned in one identification file (one sample)
+ total_spectra_used=2,///< total_spectra_used in one identification file (one sample)
+ total_peptide_used=3,///< total number of peptides generated and used in identification
+ total_proteins_used=4,///< total number of proteins generated and used in identification
+ total_unique_assigned=5,///< total number unique peptide sequence assigned
+
+};
+
/** \def GroupingType list of available grouping algoritms
*
*/
--
GitLab