From eccbbcd0e0181b42cbc6f1a90080366deb3fc6e1 Mon Sep 17 00:00:00 2001
From: Olivier Langella <olivier.langella@u-psud.fr>
Date: Mon, 9 Jul 2018 10:53:15 +0200
Subject: [PATCH] pepxml : take into account buggy engines that do not use
correctly protein accession and description
---
src/core/proteinxtp.cpp | 2 +-
src/core/proteinxtp.h | 2 ++
src/input/pepxmlsaxhandler.cpp | 27 ++++++++++++++++++++++++---
3 files changed, 27 insertions(+), 4 deletions(-)
diff --git a/src/core/proteinxtp.cpp b/src/core/proteinxtp.cpp
index c5f99aea..6a59f4e2 100644
--- a/src/core/proteinxtp.cpp
+++ b/src/core/proteinxtp.cpp
@@ -101,8 +101,8 @@ const FastaFile * ProteinXtp::getFastaFileP() const {
void ProteinXtp::setCompleteDescription(const QString & full_description) {
setAccession (full_description.simplified().section(" ", 0,0));
setDescription (full_description.simplified().section(" ", 1));
-
}
+
void ProteinXtp::setIsContaminant(bool conta) {
_is_contaminant = conta;
}
diff --git a/src/core/proteinxtp.h b/src/core/proteinxtp.h
index adb23610..a49d39eb 100644
--- a/src/core/proteinxtp.h
+++ b/src/core/proteinxtp.h
@@ -58,6 +58,8 @@ public:
ProteinXtpSp makeProteinXtpSp() const;
+ /** @brief separates accession from description based on first space separator
+ */
void setCompleteDescription(const QString & full_description);
void setIsContaminant(bool conta);
diff --git a/src/input/pepxmlsaxhandler.cpp b/src/input/pepxmlsaxhandler.cpp
index a51e5f60..bb5a0e0f 100644
--- a/src/input/pepxmlsaxhandler.cpp
+++ b/src/input/pepxmlsaxhandler.cpp
@@ -326,7 +326,7 @@ PepXmlSaxHandler::startElement_spectrum_query(QXmlAttributes attributes)
//<alternative_protein protein="sp|P46784|RS10B_YEAST" protein_descr="40S
// ribosomal protein S10-B OS=Saccharomyces cerevisiae (strain ATCC 204508
-// \
+// \
//S288c) GN=RPS10B PE=1 SV=1" num_tol_term="2" peptide_prev_aa="K"
// peptide_next_aa="N"/>
bool
@@ -335,7 +335,14 @@ PepXmlSaxHandler::startElement_alternative_protein(QXmlAttributes attributes)
bool is_ok = true;
ProteinXtpSp sp_xtp_protein = ProteinXtp().makeProteinXtpSp();
sp_xtp_protein.get()->setAccession(attributes.value("protein"));
- sp_xtp_protein.get()->setDescription(attributes.value("protein_descr"));
+ if(attributes.value("protein_descr").isEmpty())
+ {
+ sp_xtp_protein.get()->setCompleteDescription(attributes.value("protein"));
+ }
+ else
+ {
+ sp_xtp_protein.get()->setDescription(attributes.value("protein_descr"));
+ }
sp_xtp_protein.get()->setFastaFileP(
_p_identification_data_source->getFastaFileList()[0].get());
@@ -364,7 +371,14 @@ PepXmlSaxHandler::startElement_search_hit(QXmlAttributes attributes)
//_current_protein.setAccession(attributes.value("protein"));
ProteinXtpSp sp_xtp_protein = ProteinXtp().makeProteinXtpSp();
sp_xtp_protein.get()->setAccession(attributes.value("protein"));
- sp_xtp_protein.get()->setDescription(attributes.value("protein_descr"));
+ if(attributes.value("protein_descr").isEmpty())
+ {
+ sp_xtp_protein.get()->setCompleteDescription(attributes.value("protein"));
+ }
+ else
+ {
+ sp_xtp_protein.get()->setDescription(attributes.value("protein_descr"));
+ }
sp_xtp_protein.get()->setFastaFileP(
_p_identification_data_source->getFastaFileList()[0].get());
@@ -413,6 +427,13 @@ PepXmlSaxHandler::startElement_search_hit(QXmlAttributes attributes)
// search_engine="X! Tandem (k-score)"
search_engine = IdentificationEngine::XTandem;
}
+
+ else if(_current_search_engine == "X! Tandem")
+ {
+ // files coming from msfragger
+ search_engine = IdentificationEngine::XTandem;
+ }
+
else if(_current_search_engine == "OMSSA")
{
search_engine = IdentificationEngine::OMSSA;
--
GitLab