Skip to content
Snippets Groups Projects
Commit e23ededd authored by Olivier Langella's avatar Olivier Langella
Browse files

WIP : writing PROTICdbML

parent 73d0b272
No related branches found
No related tags found
No related merge requests found
Hide whitespace changes
Inline Side-by-side
src/output/proticdbml.cpp
+ 147
− 160
View file @ e23ededd
......@@ -392,10 +392,10 @@ void ProticdbMl::writeIdentificationRun(IdentificationGroup * p_identification)
//}
writeCvParam("PROTICdbO:0000325", QString("%1").arg(_sp_project.get()->getAutomaticFilterParameters().getFilterMinimumPeptidePerMatch()),
"X!TandemPipeline filter on minimal number of peptide per protein");
"X!TandemPipeline filter on minimal number of peptide per protein");
writeCvParam("PROTICdbO:0000324", QString("%1").arg(_sp_project.get()->getAutomaticFilterParameters().getFilterPeptideEvalue()),
"X!TandemPipeline filter on peptide evalue");
"X!TandemPipeline filter on peptide evalue");
// TODO write database filter
// this.writeCvParam("PROTICdbO:0000324", ""
......@@ -410,173 +410,46 @@ void ProticdbMl::writeIdentificationRun(IdentificationGroup * p_identification)
_output_stream->writeEndElement();// description
}
private void writePeptideHit(Peptide pep, Group group,
PeptideMass peptideMass) throws XMLStreamException, MSMSException {
if (peptideMass == null) {
String message = "peptideMass == null";
logger.error(message);
throw new MSMSException(message);
}
String pep_hit_id = "pep" + pepSampleScanCle2xmlId.size();
// param par default
writer.writeStartElement("peptideHit");
writer.writeAttribute("calc_mr", "" + (pep.get_mhplus_theo() - 1.00794));
double exp_mz = (pep.get_mhplus_obser() + (1.00794 * (pep.getCharge() - 1)))
/ pep.getCharge();
writer.writeAttribute("exp_mz", "" + exp_mz);
writer.writeAttribute("delta", "" + pep.get_deltamass());
writer.writeAttribute("exp_mr", "" + (pep.get_mhplus_obser() - 1.00794));
writer.writeAttribute("acq_number", "" + pep.get_scan());
writer.writeAttribute("ms_run_id",
"" + msrun_to_id.get(pep.get_sample()));
writer.writeAttribute("id", pep_hit_id);
writer.writeAttribute("exp_z", "" + pep.getCharge());
this.writeCvParam("PROTICdbO:0000339",
Utils.getPappsoPeptideMassId(group, peptideMass),
"peptide mass id");
if (xtpExperimentType.equals("phospho")) {
this.writeCvParam("PROTICdbO:0000349",
Utils.getPappsoPhosphoPeptideMassId(group, peptideMass),
"phosphopeptide mass id");
void ProticdbMl::writepeptideHits(IdentificationGroup * p_identification) {
_output_stream->writeStartElement("peptideHits");
for (std::pair <unsigned int, GroupingGroupSp > group_pair : p_identification->getGroupStore().getGroupMap()) {
writepeptideHitsbyGroup(group_pair.second);
}
// pep.getMsRun().
IdentificationDataSource idDataSource = pep
.getIdentificationDataSource();
if (idDataSource.getClass().equals(IdentificationXtandemFile.class)) {
// cvparam specifique xtandem
this.writeCvParam("PROTICdbO:0000287", "" + pep.get_evalue(),
"xtandem peptide evalue");
this.writeCvParam("PROTICdbO:0000288", "" + pep.get_hypercorr(),
"xtandem peptide hyperscore");
}
this.writeCvParam("PROTICdbO:0000289",
"" + pep.getRetentionTimeInMinutes(),
"xtandem peptide retention time");
/*
* writer.writeStartElement("cvParam"); writer.writeAttribute("value",
* "" + pep.get_RT()); writer.writeAttribute("name",
* "xtandem peptide retention time"); writer.writeAttribute("cvLabel",
* "PROTICdbO"); writer.writeAttribute("accession",
* "PROTICdbO:0000289"); writer.writeEndElement();// cvParam
*/
// Element cvParam4 = doc.createElement("cvParam");
// cvParam4.setAttribute("value", "" + pep.get_post());
// cvParam4.setAttribute("name", "xtandem peptide post");
// cvParam4.setAttribute("cvLabel", "PROTICdbO");
// cvParam4.setAttribute("accession", "PROTICdbO:00000XX");
// peptideHit.appendChild(cvParam4);
//
// Element cvParam5 = doc.createElement("cvParam");
// cvParam5.setAttribute("value", "" + pep.get_pre());
// cvParam5.setAttribute("name", "xtandem peptide pre");
// cvParam5.setAttribute("cvLabel", "PROTICdbO");
// cvParam5.setAttribute("accession", "PROTICdbO:00000XX");
// peptideHit.appendChild(cvParam5);
// sequences avec les modifs
writer.writeStartElement("pepSeq");
writer.writeStartElement("peptide");
writer.writeCharacters("" + pep.getSequence());
writer.writeEndElement();// peptide
if (pep.get_Modifs().size() != 0) {
this.writePtms(pep.get_Modifs());
}
writer.writeEndElement();// pepSeq
pepSampleScanCle2xmlId.put(pep.get_sample_scan_sequencepeptide_cle(),
pep_hit_id);
writer.writeEndElement();// peptideHit
}
private void writePtms(ArrayList<Modifs> mods) throws XMLStreamException {
writer.writeStartElement("ptms");
for (Modifs mod : mods) {
writer.writeStartElement("ptm");
writer.writeAttribute("diff_mono", "" + mod.get_modvalue());
writer.writeAttribute("position", "" + mod.get_posi());
writer.writeAttribute("aa", "" + mod.get_AA());
try {
this.writeOboModif(mod.getPsiMod());
} catch (MSMSException e) {
String message = "error writing OBO modif : " + e.getMessage();
logger.warn(message);
}
writer.writeEndElement();// ptm
}
writer.writeEndElement();// ptms
_output_stream->writeEndElement();// "peptideHits");
}
void ProticdbMl::writeOboModif(pappso::AaModificationP mod) {
_output_stream->writeStartElement("cvParam");
_output_stream->writeAttribute("name", mod->getName());
_output_stream->writeAttribute("cvLabel", "MOD");
_output_stream->writeAttribute("accession", mod->getAccession());
_output_stream->writeEndElement();// cvParam
QString getProticPeptideHitKey(PeptideMatch * p_peptide_match) {
return (QString ("%1 %2 %3").arg(p_peptide_match->getHashSampleScan()).arg(p_peptide_match->getPeptideXtpSp().get()->toAbsoluteString()).arg(p_peptide_match->getIdentificationDataSource()->getResourceName()));
}
private void writepeptideHits(Identification identification)
throws MSMSException {
try {
writer.writeStartElement("peptideHits");
for (Group group : identification.getGrouping().getGroupList()) {
// generating peptide numbers
PeptideMassSet pepMassSet = new PeptideMassSet();
struct ProticPeptideHit {
bool operator ==(const ProticPeptideHit &other) const {
return (key == other.key);
};
/** key = getProticPeptideHitKey
* */
QString key;
PeptideMatch * peptide_match;
};
Hashtable<String, PeptideMass> pepSampleScanCle2peptideMass;
if (this.xtpExperimentType.equals("phospho")) {
pepSampleScanCle2peptideMass = pepMassSet
.numberingPhosphoGetSampleScanCle2peptideMass(group);
void ProticdbMl::writepeptideHitsbyGroup(GroupingGroup * p_group) {
} else {
pepSampleScanCle2peptideMass = pepMassSet
.numberingGetSampleScanCle2peptideMass(group);
}
for (SubGroup sg : group.getSubGroupSet().getSubGroupList()) {
for (HashSampleScanSetProt hashProt : sg
.getHashSampleScanSetProtList()) {
ArrayList<Peptide> peptideList = null;
if (this.xtpExperimentType.equals("phospho")) {
peptideList = ((Match) hashProt.getGrpProtMatch())
.getValidPeptideInPhosphoIsland((HashSampleScanSetPhospho) hashProt);
} else {
peptideList = ((Match) hashProt.getGrpProtMatch())
.get_valid_peptides();
}
for (Peptide pep : peptideList) {
std::vector <ProticPeptideHit> protic_peptide_hit_list;
if (pepSampleScanCle2xmlId.containsKey(pep
.get_sample_scan_sequencepeptide_cle()) == false) {
writePeptideHit(
pep,
group,
pepSampleScanCle2peptideMass.get(pep
.get_sample_scan_sequencepeptide_cle()));
}
}
}
}
}
writer.writeEndElement();// "peptideHits");
} catch (Exception e) {
String message = "error writing peptideHits in ProticDbMl \n"
+ e.getMessage();
logger.error(message);
throw new MSMSException(message);
for (PeptideMatch* p_peptide_match : p_group->getPeptideMatchList()) {
protic_peptide_hit_list.push_back(ProticPeptideHit(getProticPeptideHitKey(p_peptide_match), p_peptide_match));
}
std::sort(protic_peptide_hit_list.begin(),protic_peptide_hit_list.end(),[](const ProticPeptideHit & first, const ProticPeptideHit & second) {
return (first.key < second.key) ;
});
auto last = std::unique(protic_peptide_hit_list.begin(), protic_peptide_hit_list.end());
protic_peptide_hit_list.erase(last, protic_peptide_hit_list.end());
for (ProticPeptideHit & protic_peptide_hit :protic_peptide_hit_list) {
QString xml_id = QString("pep%1").arg(_peptidekey_to_id.size()+1);
writePeptideHit(xml_id, protic_peptide_hit);
_peptidekey_to_id.insert(std::pair<QString, QString>(protic_peptide_hit.key, xml_id));
}
}
......@@ -748,6 +621,120 @@ throws MSMSException {
}
}
void ProticdbMl::writePeptideHit(Peptide pep, Group group,
PeptideMass peptideMass) throws XMLStreamException, MSMSException {
if (peptideMass == null) {
String message = "peptideMass == null";
logger.error(message);
throw new MSMSException(message);
}
String pep_hit_id = "pep" + pepSampleScanCle2xmlId.size();
// param par default
writer.writeStartElement("peptideHit");
writer.writeAttribute("calc_mr", "" + (pep.get_mhplus_theo() - 1.00794));
double exp_mz = (pep.get_mhplus_obser() + (1.00794 * (pep.getCharge() - 1)))
/ pep.getCharge();
writer.writeAttribute("exp_mz", "" + exp_mz);
writer.writeAttribute("delta", "" + pep.get_deltamass());
writer.writeAttribute("exp_mr", "" + (pep.get_mhplus_obser() - 1.00794));
writer.writeAttribute("acq_number", "" + pep.get_scan());
writer.writeAttribute("ms_run_id",
"" + msrun_to_id.get(pep.get_sample()));
writer.writeAttribute("id", pep_hit_id);
writer.writeAttribute("exp_z", "" + pep.getCharge());
this.writeCvParam("PROTICdbO:0000339",
Utils.getPappsoPeptideMassId(group, peptideMass),
"peptide mass id");
if (xtpExperimentType.equals("phospho")) {
this.writeCvParam("PROTICdbO:0000349",
Utils.getPappsoPhosphoPeptideMassId(group, peptideMass),
"phosphopeptide mass id");
}
// pep.getMsRun().
IdentificationDataSource idDataSource = pep
.getIdentificationDataSource();
if (idDataSource.getClass().equals(IdentificationXtandemFile.class)) {
// cvparam specifique xtandem
this.writeCvParam("PROTICdbO:0000287", "" + pep.get_evalue(),
"xtandem peptide evalue");
this.writeCvParam("PROTICdbO:0000288", "" + pep.get_hypercorr(),
"xtandem peptide hyperscore");
}
this.writeCvParam("PROTICdbO:0000289",
"" + pep.getRetentionTimeInMinutes(),
"xtandem peptide retention time");
/*
* writer.writeStartElement("cvParam"); writer.writeAttribute("value",
* "" + pep.get_RT()); writer.writeAttribute("name",
* "xtandem peptide retention time"); writer.writeAttribute("cvLabel",
* "PROTICdbO"); writer.writeAttribute("accession",
* "PROTICdbO:0000289"); writer.writeEndElement();// cvParam
*/
// Element cvParam4 = doc.createElement("cvParam");
// cvParam4.setAttribute("value", "" + pep.get_post());
// cvParam4.setAttribute("name", "xtandem peptide post");
// cvParam4.setAttribute("cvLabel", "PROTICdbO");
// cvParam4.setAttribute("accession", "PROTICdbO:00000XX");
// peptideHit.appendChild(cvParam4);
//
// Element cvParam5 = doc.createElement("cvParam");
// cvParam5.setAttribute("value", "" + pep.get_pre());
// cvParam5.setAttribute("name", "xtandem peptide pre");
// cvParam5.setAttribute("cvLabel", "PROTICdbO");
// cvParam5.setAttribute("accession", "PROTICdbO:00000XX");
// peptideHit.appendChild(cvParam5);
// sequences avec les modifs
writer.writeStartElement("pepSeq");
writer.writeStartElement("peptide");
writer.writeCharacters("" + pep.getSequence());
writer.writeEndElement();// peptide
if (pep.get_Modifs().size() != 0) {
this.writePtms(pep.get_Modifs());
}
writer.writeEndElement();// pepSeq
pepSampleScanCle2xmlId.put(pep.get_sample_scan_sequencepeptide_cle(),
pep_hit_id);
writer.writeEndElement();// peptideHit
}
private void writePtms(ArrayList<Modifs> mods) throws XMLStreamException {
writer.writeStartElement("ptms");
for (Modifs mod : mods) {
writer.writeStartElement("ptm");
writer.writeAttribute("diff_mono", "" + mod.get_modvalue());
writer.writeAttribute("position", "" + mod.get_posi());
writer.writeAttribute("aa", "" + mod.get_AA());
try {
this.writeOboModif(mod.getPsiMod());
} catch (MSMSException e) {
String message = "error writing OBO modif : " + e.getMessage();
logger.warn(message);
}
writer.writeEndElement();// ptm
}
writer.writeEndElement();// ptms
}
void ProticdbMl::writeOboModif(pappso::AaModificationP mod) {
_output_stream->writeStartElement("cvParam");
_output_stream->writeAttribute("name", mod->getName());
_output_stream->writeAttribute("cvLabel", "MOD");
_output_stream->writeAttribute("accession", mod->getAccession());
_output_stream->writeEndElement();// cvParam
}
void ProticdbMl::writeCvParam(QString acc, QString value, QString description) {
_output_stream->writeStartElement("cvParam");
......
src/output/proticdbml.h
+ 3
− 0
View file @ e23ededd
......@@ -57,6 +57,8 @@ private :
void writeMsRuns();
void writeIdentificationRun(IdentificationGroup * p_identification);
void writeIdentificationDataSource(IdentificationDataSource * p_identification_data_source);
void writepeptideHits(IdentificationGroup * p_identification);
void writepeptideHitsbyGroup(GroupingGroup * p_group);
private :
QFile * _output_file;
......@@ -66,6 +68,7 @@ private :
QTime _duracel;
std::map<QString, QString> _map_accession2xmlid;
std::map<QString, QString> _sample_to_id;
std::map<QString, QString> _peptidekey_to_id;
};
#endif // PROTICDBML_H
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment