From 2c3eb033e78e5ff6a6a90f9d3d5ba8c1e7d328a7 Mon Sep 17 00:00:00 2001
From: Olivier Langella <olivier.langella@u-psud.fr>
Date: Mon, 25 Sep 2017 21:52:43 +0200
Subject: [PATCH] introducing MsRunStatistics
---
src/core/msrun.cpp | 12 +++++++++
src/core/msrun.h | 14 +++++++++++
src/output/ods/samplesheet.cpp | 45 ++++++++++++++++++++++++++++------
src/utils/readspectrum.cpp | 23 +++++++++++++++--
src/utils/readspectrum.h | 4 +++
src/utils/types.h | 13 +++++++++-
6 files changed, 100 insertions(+), 11 deletions(-)
diff --git a/src/core/msrun.cpp b/src/core/msrun.cpp
index cb6b940ec..390f27a0d 100644
--- a/src/core/msrun.cpp
+++ b/src/core/msrun.cpp
@@ -82,6 +82,18 @@ MzFormat MsRun::getMzFormat() const {
return _mz_format;
}
+void MsRun::setMsRunStatistics(MsRunStatistics param, const QVariant& value) {
+ _param_stats.insert(std::pair<MsRunStatistics, QVariant>(param, value));
+}
+const QVariant MsRun::getMsRunStatistics(MsRunStatistics param) const {
+ try {
+ return _param_stats.at(param);
+ }
+ catch (std::out_of_range) {
+ return QVariant();
+ }
+}
+
/*
void MsRun::setMzFormat(MzFormat mz_format) {
_mz_format = mz_format;
diff --git a/src/core/msrun.h b/src/core/msrun.h
index a10603c98..ce038fc8d 100644
--- a/src/core/msrun.h
+++ b/src/core/msrun.h
@@ -34,6 +34,7 @@
#include <memory>
#include <QString>
+#include <QVariant>
#include "../utils/types.h"
class MsRun;
@@ -55,11 +56,24 @@ public:
MzFormat getMzFormat() const;
//void setMzFormat(MzFormat mz_format);
+
+
+ /** \brief set MS run statistics
+ * any statistics on this MS run file
+ */
+ virtual void setMsRunStatistics(MsRunStatistics param, const QVariant& value);
+
+
+ /** \brief get MS run statistics
+ */
+ virtual const QVariant getMsRunStatistics(MsRunStatistics param) const;
+
private :
QString _xml_id;
QString _location;
QString _name;
MzFormat _mz_format=MzFormat::mzXML;
+ std::map<MsRunStatistics, QVariant> _param_stats;
};
#endif // MSRUN_H
diff --git a/src/output/ods/samplesheet.cpp b/src/output/ods/samplesheet.cpp
index 2a406f717..170cae5c8 100644
--- a/src/output/ods/samplesheet.cpp
+++ b/src/output/ods/samplesheet.cpp
@@ -28,6 +28,7 @@
******************************************************************************/
#include "samplesheet.h"
+#include "../../utils/readspectrum.h"
SampleSheet::SampleSheet (OdsExport * p_ods_export, CalcWriterInterface * p_writer, const Project * p_project): _p_project(p_project) {
_p_writer = p_writer;
@@ -88,9 +89,12 @@ void SampleSheet::writeHeaders() {
void SampleSheet::writeIdentificationDataSource(IdentificationDataSource * p_ident_data_source) {
_p_writer->writeLine();
- _p_writer->writeCell(p_ident_data_source->getMsRunSp().get()->getXmlId());
- _p_writer->writeCell(p_ident_data_source->getMsRunSp().get()->getSampleName());
- _p_writer->writeCell(p_ident_data_source->getMsRunSp().get()->getFilename());
+ MsRunSp msrun_sp = p_ident_data_source->getMsRunSp();
+ SpectrumStore::checkMsRunStatistics(msrun_sp);
+
+ _p_writer->writeCell(msrun_sp.get()->getXmlId());
+ _p_writer->writeCell(msrun_sp.get()->getSampleName());
+ _p_writer->writeCell(msrun_sp.get()->getFilename());
_p_writer->writeCell(p_ident_data_source->getIdentificationEngineName());
_p_writer->writeCell(p_ident_data_source->getIdentificationEngineVersion());
_p_writer->writeCell(p_ident_data_source->getResourceName());
@@ -101,15 +105,40 @@ 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());
+ QVariant var_sample = p_ident_data_source->getIdentificationEngineParam(IdentificationEngineParam::tandem_param);
+ if (var_sample.isNull()) {
+ _p_writer->writeEmptyCell();
+ } else {
+ _p_writer->writeCell(var_sample.toString());
+ }
+ var_sample = p_ident_data_source->getIdentificationEngineStatistics(IdentificationEngineStatistics::total_spectra_assigned);
+ if (var_sample.isNull()) {
+ _p_writer->writeEmptyCell();
+ } else {
+ _p_writer->writeCell(var_sample.toInt());
+ }
+ var_sample = p_ident_data_source->getIdentificationEngineStatistics(IdentificationEngineStatistics::total_spectra_used);
+ if (var_sample.isNull()) {
+ _p_writer->writeEmptyCell();
+ } else {
+ _p_writer->writeCell(var_sample.toInt());
+ }
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());
}
+ else {
+ _p_writer->writeEmptyCell();
+ }
+
- _p_writer->writeCell(p_ident_data_source->getIdentificationEngineStatistics(IdentificationEngineStatistics::total_unique_assigned).toString());
+ var_sample = p_ident_data_source->getIdentificationEngineStatistics(IdentificationEngineStatistics::total_unique_assigned);
+ if (var_sample.isNull()) {
+ _p_writer->writeEmptyCell();
+ } else {
+ _p_writer->writeCell(var_sample.toInt());
+ }
+
+
}
diff --git a/src/utils/readspectrum.cpp b/src/utils/readspectrum.cpp
index b84e14943..67c8579ce 100644
--- a/src/utils/readspectrum.cpp
+++ b/src/utils/readspectrum.cpp
@@ -258,8 +258,20 @@ const QString SpectrumStore::findMzFile(const QString &filename) {
return QString();
}
-pappso::SpectrumSp SpectrumStore::getSpectrumSpFromMsRunSp(MsRunSp msrun, unsigned int scan_num) {
- pappso::QualifiedSpectrum spectrum;
+void SpectrumStore::checkMsRunStatistics(MsRunSp msrun) {
+ QVariant msrun_var = msrun.get()->getMsRunStatistics(MsRunStatistics::total_spectra);
+ if (msrun_var.isNull()) {
+ try {
+ pwiz::msdata::MSDataFile * p_msdatafile = findPwizMSDataFile(msrun);
+ msrun.get()->setMsRunStatistics(MsRunStatistics::total_spectra, (unsigned int) p_msdatafile->run.spectrumListPtr.get()->size());
+ }
+ catch (pappso::ExceptionNotFound& error) {
+ //no file found, no statistics
+ }
+ }
+}
+
+pwiz::msdata::MSDataFile * SpectrumStore::findPwizMSDataFile(MsRunSp msrun) {
pwiz::msdata::MSDataFile * p_msdatafile = nullptr;
std::map<MsRunSp, pwiz::msdata::MSDataFile *>::iterator it_msdata = _map_msrun_msdatafile.find(msrun);
if (it_msdata == _map_msrun_msdatafile.end()) {
@@ -279,6 +291,13 @@ pappso::SpectrumSp SpectrumStore::getSpectrumSpFromMsRunSp(MsRunSp msrun, unsign
else {
p_msdatafile =it_msdata->second;
}
+ return p_msdatafile;
+}
+
+pappso::SpectrumSp SpectrumStore::getSpectrumSpFromMsRunSp(MsRunSp msrun, unsigned int scan_num) {
+ pappso::QualifiedSpectrum spectrum;
+ pwiz::msdata::MSDataFile * p_msdatafile = findPwizMSDataFile(msrun);
+
if (p_msdatafile == nullptr) {
return spectrum.getOriginalSpectrumSp();
}
diff --git a/src/utils/readspectrum.h b/src/utils/readspectrum.h
index f8538bede..8051910c4 100644
--- a/src/utils/readspectrum.h
+++ b/src/utils/readspectrum.h
@@ -42,8 +42,12 @@ class MSDataFile;
class SpectrumStore {
public:
static pappso::SpectrumSp getSpectrumSpFromMsRunSp(MsRunSp msrun, unsigned int scan_num);
+ /** @brief try to find MS run statistics
+ */
+ static void checkMsRunStatistics(MsRunSp msrun);
private:
static const QString findMzFile(const QString &filename);
+ static pwiz::msdata::MSDataFile * findPwizMSDataFile(MsRunSp msrun) ;
private:
static std::map<MsRunSp, pwiz::msdata::MSDataFile *> _map_msrun_msdatafile;
};
diff --git a/src/utils/types.h b/src/utils/types.h
index a264e3aaf..619d3b7ce 100644
--- a/src/utils/types.h
+++ b/src/utils/types.h
@@ -74,9 +74,20 @@ enum class IdentificationEngineStatistics: std::int8_t {
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 MsRunStatistics MS run statistics
+ *
+ */
+enum class MsRunStatistics: std::int8_t {
+ total_spectra=1, ///< total number of spectra
+ total_spectra_ms1=2,///< total number of MS1 spectra
+ total_spectra_ms2=3,///< total number of MS2 spectra
+ total_spectra_ms3=4,///< total number of MS3 spectra
+};
+
+
+
/** \def GroupingType list of available grouping algoritms
*
*/
--
GitLab