From b530a62bcf8009004a65b337254072cbc859d234 Mon Sep 17 00:00:00 2001
From: Olivier Langella <Olivier.Langella@moulon.inra.fr>
Date: Tue, 28 Mar 2017 15:07:59 +0200
Subject: [PATCH] PAI computation
---
src/core/proteinmatch.cpp | 32 ++++++++++++++++++-
src/core/proteinmatch.h | 12 +++++++
src/core/proteinxtp.cpp | 30 +++++++++++++++++
src/core/proteinxtp.h | 1 +
src/gui/project_view/projectwindow.cpp | 10 +++---
.../protein_list_view/proteintablemodel.cpp | 7 +++-
6 files changed, 85 insertions(+), 7 deletions(-)
diff --git a/src/core/proteinmatch.cpp b/src/core/proteinmatch.cpp
index ae009c04..9df85654 100644
--- a/src/core/proteinmatch.cpp
+++ b/src/core/proteinmatch.cpp
@@ -162,6 +162,36 @@ size_t ProteinMatch::countUniqueSequence()const {
return sequence_list.size();
}
+unsigned int ProteinMatch::countValidAndCheckedPeptideMassCharge(const pappso::MsRunIdSp & sp_msrun_id) const {
+ std::set<QString> sequence_list;
+ for (auto & p_peptide_match : _peptide_match_list) {
+ if (p_peptide_match->isValidAndChecked()) {
+ if(sp_msrun_id.get() != nullptr) {
+ //within sample
+ if (p_peptide_match->getMsRunIdSp().get() == sp_msrun_id.get()) {
+ sequence_list.insert(QString("%1-%2").arg(p_peptide_match->getPeptideXtpSp().get()->toAbsoluteString()).arg(p_peptide_match->getCharge()));
+ }
+ }
+ else {
+ //overall samples
+ sequence_list.insert(QString("%1-%2").arg(p_peptide_match->getPeptideXtpSp().get()->toAbsoluteString()).arg(p_peptide_match->getCharge()));
+ }
+ }
+ }
+ return sequence_list.size();
+}
+
+
+pappso::pappso_double ProteinMatch::getPAI() const {
+ pappso::MsRunIdSp sp_msrun_id;
+ return getPAI(sp_msrun_id);
+}
+
+pappso::pappso_double ProteinMatch::getPAI(const pappso::MsRunIdSp & sp_msrun_id) const {
+ pappso::pappso_double PAI = (pappso::pappso_double) countValidAndCheckedPeptideMassCharge(sp_msrun_id) / (pappso::pappso_double) _protein_sp.get()->countTrypticPeptidesForPAI();
+ return PAI;
+}
+
pappso::pappso_double ProteinMatch::getCoverage() const {
size_t prot_size = _protein_sp.get()->size();
//qDebug() << "ProteinMatch::getCoverage begin prot_size=" << prot_size << " " << _protein_sp.get()-//>getSequence();
@@ -186,7 +216,7 @@ pappso::pappso_double ProteinMatch::getCoverage() const {
return (((pappso::pappso_double)count)/ ((pappso::pappso_double)prot_size));
}
const GroupingGroupSp & ProteinMatch::getGroupingGroupSp() const {
- return _sp_group;
+ return _sp_group;
}
void ProteinMatch::setGroupInstance(GroupStore & group_store) {
diff --git a/src/core/proteinmatch.h b/src/core/proteinmatch.h
index b473d43e..d649e355 100644
--- a/src/core/proteinmatch.h
+++ b/src/core/proteinmatch.h
@@ -46,6 +46,17 @@ public:
void setEvalue(pappso::pappso_double evalue);
pappso::pappso_double getEvalue() const;
pappso::pappso_double getCoverage() const;
+
+ /** @brief compute Protein Abundance Index (PAI)
+ * overall sample PAI computation (Rappsilber et al. 2002)
+ * */
+ pappso::pappso_double getPAI() const;
+
+ /** @brief compute Protein Abundance Index (PAI) within sample
+ * PAI computation (Rappsilber et al. 2002)
+ * */
+ pappso::pappso_double getPAI(const pappso::MsRunIdSp & sp_msrun_id) const;
+
void setProteinXtpSp(ProteinXtpSp protein_sp);
void addPeptideMatch(PeptideMatch * peptide_match);
std::vector<PeptideMatch *> & getPeptideMatchList();
@@ -95,6 +106,7 @@ private:
/** @brief automatic filter result (false by default)
*/
bool _proxy_valid = false;
+ unsigned int countValidAndCheckedPeptideMassCharge(const pappso::MsRunIdSp & sp_msrun_id) const;
};
diff --git a/src/core/proteinxtp.cpp b/src/core/proteinxtp.cpp
index 432318eb..f08e97e2 100644
--- a/src/core/proteinxtp.cpp
+++ b/src/core/proteinxtp.cpp
@@ -22,6 +22,7 @@
******************************************************************************/
#include "proteinxtp.h"
+#include <pappsomspp/protein/enzyme.h>
ProteinXtp::ProteinXtp():pappso::Protein()
{
@@ -62,3 +63,32 @@ bool ProteinXtp::isContaminant() const {
bool ProteinXtp::isDecoy() const {
return _is_decoy;
}
+
+class DigestionHandler: public pappso::EnzymeProductInterface {
+public:
+ void setPeptide(std::int8_t sequence_database_id, const pappso::ProteinSp & protein_sp, bool is_decoy, const QString& peptide, unsigned int start, bool is_nter, unsigned int missed_cleavage_number, bool semi_enzyme) override {
+ _peptide_list.append(peptide);
+ };
+
+ QStringList _peptide_list;
+};
+
+unsigned int ProteinXtp::countTrypticPeptidesForPAI() const {
+ qDebug() << "ProteinXtp::countTrypticPeptidesForPAI begin";
+ pappso::Enzyme kinase;
+ kinase.setMiscleavage(0);
+ DigestionHandler digestion;
+ kinase.eat(0,this->makeProteinSp(),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 ++;
+ }
+ }
+
+ return count;
+ qDebug() << "ProteinXtp::countTrypticPeptidesForPAI end";
+}
diff --git a/src/core/proteinxtp.h b/src/core/proteinxtp.h
index 764266d3..0dfa8ff7 100644
--- a/src/core/proteinxtp.h
+++ b/src/core/proteinxtp.h
@@ -48,6 +48,7 @@ public:
void setIsDecoy(bool conta);
bool isContaminant() const;
bool isDecoy() const;
+ unsigned int countTrypticPeptidesForPAI() const;
private:
SequenceDatabase * _p_sequence_database;
diff --git a/src/gui/project_view/projectwindow.cpp b/src/gui/project_view/projectwindow.cpp
index 96f0793d..c5700f62 100644
--- a/src/gui/project_view/projectwindow.cpp
+++ b/src/gui/project_view/projectwindow.cpp
@@ -167,17 +167,17 @@ void ProjectWindow::doViewProteinList(IdentificationGroup* p_identification_grou
void ProjectWindow::setProjectSp(ProjectSp project_sp) {
for (auto && p_window :_peptide_list_window_collection) {
- delete p_window;
+ delete p_window;
}
_peptide_list_window_collection.clear();
_p_current_peptide_list_window = nullptr;
-
- for (auto && p_window :_protein_list_window_collection) {
- delete p_window;
+
+ for (auto && p_window :_protein_list_window_collection) {
+ delete p_window;
}
_protein_list_window_collection.clear();
_p_current_protein_list_window = nullptr;
-
+
_project_sp = project_sp;
vector< MsRunIdSp > ms_run_list = _project_sp.get()->getCurrentIdentificationGroupP()->getMsRunIdSpList();
ui->sample_number_display->setText(QString("%1").arg(ms_run_list.size()));
diff --git a/src/gui/protein_list_view/proteintablemodel.cpp b/src/gui/protein_list_view/proteintablemodel.cpp
index eb74f3cd..3a64371f 100644
--- a/src/gui/protein_list_view/proteintablemodel.cpp
+++ b/src/gui/protein_list_view/proteintablemodel.cpp
@@ -147,7 +147,7 @@ int ProteinTableModel::rowCount(const QModelIndex &parent ) const {
return 0;
}
int ProteinTableModel::columnCount(const QModelIndex &parent ) const {
- return 10;
+ return 11;
}
QVariant ProteinTableModel::headerData(int section, Qt::Orientation orientation, int role) const
{
@@ -176,6 +176,8 @@ QVariant ProteinTableModel::headerData(int section, Qt::Orientation orientation,
return QString("specific sequence");
case 9:
return QString("coverage");
+ case 10:
+ return QString("PAI");
}
}
}
@@ -264,6 +266,9 @@ QVariant ProteinTableModel::data(const QModelIndex &index, int role ) const {
if (col == 9) {
return QVariant ((qreal)_p_identification_group->getProteinMatchList().at(row)->getCoverage());
}
+ if (col == 10) {
+ return QVariant ((qreal)_p_identification_group->getProteinMatchList().at(row)->getPAI());
+ }
return QVariant();
}
--
GitLab