From 73cf85856273052fcbf4f90b4badd3dfcf39fa75 Mon Sep 17 00:00:00 2001
From: Olivier Langella <Olivier.Langella@moulon.inra.fr>
Date: Fri, 7 Apr 2017 12:14:15 +0200
Subject: [PATCH] WIP : writing MassChroqML
---
src/grouping/groupinggroup.cpp | 37 +++++++++------
src/grouping/groupinggroup.h | 4 ++
src/output/masschroqml.cpp | 84 +++++++++++++++++++++-------------
src/output/masschroqml.h | 2 +
4 files changed, 79 insertions(+), 48 deletions(-)
diff --git a/src/grouping/groupinggroup.cpp b/src/grouping/groupinggroup.cpp
index b414e320..7dd05b0a 100644
--- a/src/grouping/groupinggroup.cpp
+++ b/src/grouping/groupinggroup.cpp
@@ -37,26 +37,33 @@ GroupingGroup::~GroupingGroup()
{
}
+
+const std::vector<std::pair<unsigned int, const PeptideMatch *>> & GroupingGroup::getPairSgNumberPeptideMatchList() const {
+ return _pair_sg_number_peptide_match_list;
+}
+unsigned int GroupingGroup::getGroupNumber() const {
+ return _group_number;
+}
unsigned int GroupingGroup::getNumberOfSubgroups() const {
return _number_of_subgroup;
}
const QStringList GroupingGroup::getSubgroupIdList(const PeptideMatch * p_peptide_match) const {
- std::set<unsigned int> subgroup_list;
+ std::set<unsigned int> subgroup_list;
pappso::GrpPeptide * p_grp_peptide = p_peptide_match->getGrpPeptideSp().get();
if (p_grp_peptide != nullptr) {
- for (auto && pair_peptide_match :_pair_sg_number_peptide_match_list) {
- if(pair_peptide_match.second->getGrpPeptideSp().get() == p_grp_peptide) {
- subgroup_list.insert(pair_peptide_match.first);
+ for (auto && pair_peptide_match :_pair_sg_number_peptide_match_list) {
+ if(pair_peptide_match.second->getGrpPeptideSp().get() == p_grp_peptide) {
+ subgroup_list.insert(pair_peptide_match.first);
+ }
}
}
+
+
+ QStringList sg_str_list;
+ for (unsigned int sgnum : subgroup_list) {
+ sg_str_list << QString("%1%2").arg(pappso::Utils::getLexicalOrderedString(_group_number)).arg(pappso::Utils::getLexicalOrderedString(sgnum));
}
-
-
- QStringList sg_str_list;
- for (unsigned int sgnum : subgroup_list) {
- sg_str_list << QString("%1%2").arg(pappso::Utils::getLexicalOrderedString(_group_number)).arg(pappso::Utils::getLexicalOrderedString(sgnum));
- }
return sg_str_list;
}
@@ -65,12 +72,12 @@ unsigned int GroupingGroup::countSubgroupPresence(const PeptideMatch * p_peptide
std::set<unsigned int> subgroup_list;
pappso::GrpPeptide * p_grp_peptide = p_peptide_match->getGrpPeptideSp().get();
if (p_grp_peptide != nullptr) {
- for (auto && pair_peptide_match :_pair_sg_number_peptide_match_list) {
- if(pair_peptide_match.second->getGrpPeptideSp().get() == p_grp_peptide) {
- subgroup_list.insert(pair_peptide_match.first);
+ for (auto && pair_peptide_match :_pair_sg_number_peptide_match_list) {
+ if(pair_peptide_match.second->getGrpPeptideSp().get() == p_grp_peptide) {
+ subgroup_list.insert(pair_peptide_match.first);
+ }
}
}
- }
return subgroup_list.size();
}
@@ -131,7 +138,7 @@ std::size_t GroupingGroup::countSpecificSequence(const ProteinMatch * p_protein_
}
void GroupingGroup::add(const ProteinMatch * p_protein_match) {
-_group_number = p_protein_match->getGrpProteinSp().get()->getGroupNumber();
+ _group_number = p_protein_match->getGrpProteinSp().get()->getGroupNumber();
if (p_protein_match->getGrpProteinSp().get()->getRank() == 1) {
_number_of_subgroup++;
unsigned int sg_number = p_protein_match->getGrpProteinSp().get()->getSubGroupNumber();
diff --git a/src/grouping/groupinggroup.h b/src/grouping/groupinggroup.h
index 9a48d76d..5d740da7 100644
--- a/src/grouping/groupinggroup.h
+++ b/src/grouping/groupinggroup.h
@@ -40,6 +40,8 @@ public:
GroupingGroup();
~GroupingGroup();
+
+ unsigned int getGroupNumber() const ;
void add(const ProteinMatch * p_protein_match);
std::size_t countSpecificSpectrum(const ProteinMatch * p_protein_match) const;
@@ -55,6 +57,8 @@ public:
* */
const QStringList getSubgroupIdList(const PeptideMatch * p_peptide_match) const;
+
+ const std::vector<std::pair<unsigned int, const PeptideMatch *>> & getPairSgNumberPeptideMatchList() const;
private :
unsigned int _group_number=0;
diff --git a/src/output/masschroqml.cpp b/src/output/masschroqml.cpp
index b4492759..0e17ce3c 100644
--- a/src/output/masschroqml.cpp
+++ b/src/output/masschroqml.cpp
@@ -72,7 +72,7 @@ void MassChroQml::write(ProjectSp sp_project) {
if (_sp_project.get() == nullptr) {
throw pappso::PappsoException(QObject::tr("Error writing MassChroqML file :\n project is empty"));
}
-
+
//<masschroq>
_output_stream->writeStartElement("masschroq");
_output_stream->writeAttribute("type","input");
@@ -116,27 +116,44 @@ void MassChroQml::write(ProjectSp sp_project) {
void MassChroQml::writePeptideList() {
//<peptide_list>
_output_stream->writeStartElement("peptide_list");
- /*
- <peptide id="pep0" mh="1463.626" mods="114.08" prot_ids="P1.1"
- seq="TCVADESHAGCEK">
- <modifications><!-- this tag is optional but gives an exact mass computation -->
- <psimod at="2" acc="MOD:00397"></psimod>
- <psimod at="11" acc="MOD:00397"></psimod>
- </modifications>
- <observed_in data="samp0" scan="655" z="2" />
- <observed_in data="samp1" scan="798" z="2" />
- </peptide>
- <peptide id="pep1" mh="1103.461" mods="57.04" prot_ids="P1.1"
- seq="ADESHAGCEK">
- <modifications><!-- this tag is optional but gives an exact mass computation -->
- <psimod at="8" acc="MOD:00397"></psimod>
- </modifications>
- <observed_in data="samp3" scan="663" z="2" />
- </peptide>*/
- //</peptide_list>
+ QStringList empty;
+ for (const GroupingGroupSp & group:_group_list) {
+ std::map<const PeptideMatch *, QStringList> peptide_sg_map;
+ unsigned int group_number = group.get()->getGroupNumber();
+ for (auto && pair_peptide_match :group.get()->getPairSgNumberPeptideMatchList()) {
+ unsigned int sg_number = pair_peptide_match.first;
+ const PeptideMatch * p_peptide_match = pair_peptide_match.second;
+ if (!p_peptide_match->isGrouped()) continue;
+
+ std::pair<std::map< const PeptideMatch *, QStringList>::iterator,bool> ret = peptide_sg_map.insert(std::pair<const PeptideMatch *, QStringList>(p_peptide_match, empty));
+
+ QString sg_str = QString("%1.%2.a1").arg(pappso::Utils::getLexicalOrderedString(group_number)).arg(pappso::Utils::getLexicalOrderedString(sg_number));
+ if (!ret.first->second.contains(sg_str)) {
+ ret.first->second << sg_str;
+ }
+ }
+
+ for (auto && pair_peptide_sgid : peptide_sg_map) {
+ //<peptide id="pep0" mh="1463.626" mods="114.08" prot_ids="P1.1"
+ // seq="TCVADESHAGCEK">
+ _output_stream->writeStartElement("peptide");
+ _output_stream->writeAttribute("id",pair_peptide_sgid.first->getGrpPeptideSp().get()->getGroupingId());
+ _output_stream->writeAttribute("prot_ids",pair_peptide_sgid.second.join(" "));
+ /*
+ <modifications><!-- this tag is optional but gives an exact mass computation -->
+ <psimod at="2" acc="MOD:00397"></psimod>
+ <psimod at="11" acc="MOD:00397"></psimod>
+ </modifications>
+ <observed_in data="samp0" scan="655" z="2" />
+ <observed_in data="samp1" scan="798" z="2" />*/
+ //</peptide>
+ _output_stream->writeEndElement();
+ }
+ }
+ //</peptide_list>
_output_stream->writeEndElement();
}
-
+
void MassChroQml::writeGroups() {
//<groups>
_output_stream->writeStartElement("groups");
@@ -168,21 +185,22 @@ void MassChroQml::writeProteinList() {
_p_identification_group = identification_list[0];
for (ProteinMatch * p_protein_match :_p_identification_group->getProteinMatchList()) {
if (!p_protein_match->isGrouped()) continue;
- //<protein desc="conta|P02769|ALBU_BOVIN SERUM ALBUMIN PRECURSOR."
- // id="P1.1" />
+ //<protein desc="conta|P02769|ALBU_BOVIN SERUM ALBUMIN PRECURSOR."
+ // id="P1.1" />
+ _group_list.insert(p_protein_match->getGroupingGroupSp());
pappso::GrpProteinSp grp_protein = p_protein_match->getGrpProteinSp();
if (grp_protein.get()->getRank() == 1) {
- _output_stream->writeStartElement("protein");
- _output_stream->writeAttribute("id",grp_protein.get()->getGroupingId());
- QStringList list;
- list << p_protein_match->getProteinXtpSp().get()->getAccession();
- list << p_protein_match->getProteinXtpSp().get()->getDescription();
- _output_stream->writeAttribute("desc",list.join(" "));
- //<protein desc="conta|P02770|ALBU_RAT SERUM ALBUMIN PRECURSOR."
- // id="P1.2" />
- _output_stream->writeEndElement();
+ _output_stream->writeStartElement("protein");
+ _output_stream->writeAttribute("id",grp_protein.get()->getGroupingId());
+ QStringList list;
+ list << p_protein_match->getProteinXtpSp().get()->getAccession();
+ list << p_protein_match->getProteinXtpSp().get()->getDescription();
+ _output_stream->writeAttribute("desc",list.join(" "));
+ //<protein desc="conta|P02770|ALBU_RAT SERUM ALBUMIN PRECURSOR."
+ // id="P1.2" />
+ _output_stream->writeEndElement();
}
}
- //</protein_list>
- _output_stream->writeEndElement();
+ //</protein_list>
+ _output_stream->writeEndElement();
}
diff --git a/src/output/masschroqml.h b/src/output/masschroqml.h
index 47559be0..031695c3 100644
--- a/src/output/masschroqml.h
+++ b/src/output/masschroqml.h
@@ -34,6 +34,7 @@
#include <QFile>
#include <QString>
#include "../core/project.h"
+#include "../grouping/groupinggroup.h"
class MassChroQml
{
@@ -52,6 +53,7 @@ private :
QXmlStreamWriter * _output_stream;
ProjectSp _sp_project;
IdentificationGroup * _p_identification_group;
+ std::set<GroupingGroupSp> _group_list;
};
#endif // MASSCHROQML_H
--
GitLab