#include <mzidentmlreader.h>

Inheritance diagram for pappso::cbor::psm::MzIdentMlReader:

Classes
struct	CvParam
struct	Modification
struct	MzidDBSequence
struct	MzidSearchDatabase
struct	MzidSpectraData
struct	MzidPeptideEvidence
struct	UserParam
struct	SpectrumIdentificationItem
struct	SpectrumIdentificationResult

Public Member Functions
	MzIdentMlReader (pappso::UiMonitorInterface p_monitor, pappso::cbor::CborStreamWriter p_output, const QFileInfo &mzident_file)
virtual	~MzIdentMlReader ()
Public Member Functions inherited from pappso::XmlStreamReaderInterface
	XmlStreamReaderInterface ()
virtual	~XmlStreamReaderInterface ()
virtual bool	readFile (const QString &fileName)
bool	read (QIODevice *device)
bool	read (const QString &xml_content)
QString	errorString () const

Protected Member Functions
virtual void	readStream () override
Protected Member Functions inherited from pappso::XmlStreamReaderInterface
void	cloneStartElement (QXmlStreamWriter &output) const
void	cloneElement (QXmlStreamWriter &output)
void	cloneNode (QXmlStreamWriter &output)

Private Types
enum class	IdentificationEngine : std::int8_t { unknown = 0 , XTandem = 1 , mascot = 2 , peptider = 3 , OMSSA = 4 , SEQUEST = 5 , Comet = 6 , Morpheus = 7 , MSGFplus = 8 , SpecOMS = 9 , sage = 10 , PEAKS_Studio = 11 }

Private Member Functions
bool	readAnalysisSoftware ()
bool	readSequenceCollectionItem ()
bool	readAnalysisCollectionItem ()
bool	readDataCollectionItem ()
void	readDBSequence ()
void	readPeptide ()
bool	readPeptideEvidence ()
void	readSpectrumIdentification ()
void	readInputs ()
void	readAnalysisData ()
bool	readSearchDatabase ()
void	readSpectraData ()
void	readSpectrumIdentificationResult ()
void	readAnalysisProtocolCollection ()
void	finalDebrief ()
CvParam	readCvParam ()
UserParam	readUserParam ()
void	readSpectrumIdentificationItem (SpectrumIdentificationResult &spectrum_identification_result)
void	writeSpectrumIdentificationResult (const SpectrumIdentificationResult &spectrum_identificatio_result)
bool	writeTandemEval (const std::vector< CvParam > &cv_param_list)
void	writeSpectrumIdentificationItem (const SpectrumIdentificationItem &spectrum_identification_item)

Private Attributes
pappso::UiMonitorInterface *	mp_monitor
IdentificationEngine	m_identificationEngine
QString	m_analysisSoftwareVersion
pappso::cbor::CborStreamWriter *	mp_cborWriter
PsmProteinMap	m_proteinMap
std::map< QString, IdentificationEngine >	m_IdentificationEngineMap
	store association between xml ID and an identification engine
std::map< QString, MzidSearchDatabase >	m_mzidSearchDatabaseIdMap
	store association between xml ID and fasta files
std::map< QString, PeptideSp >	m_PeptideIdMap
	store association between xml ID and peptide sequence
std::map< QString, MzidPeptideEvidence >	m_MzidPeptideEvidenceIdMap
	store association between xml ID and peptide evidence
std::map< QString, MzidSpectraData >	m_mzidSpectraDataIdMap
	store association between xml ID and SpectraData
std::map< QString, MzidDBSequence >	m_MzidDBSequenceIdMap
	store association between xml ID and DBSequence
std::map< QString, std::vector< ProteinSp > >	m_searchDatabase_ref2proteinList
	associates database ref id to protein shared pointer because the search database id is not described before the protein (silly idea IMHO) we keep association of protein to database in this map until the real search database definition appears We then have to reprocess each protein to set the right fasta file pointer
std::map< QString, std::vector< SpectrumIdentificationResult > >	m_spectrumIdentificationResultBySpectraIdMap
	store all identification results by spectra xml id
QFileInfo	m_mzidentFile
QCborMap	m_cborParameterMap

Additional Inherited Members
Protected Attributes inherited from pappso::XmlStreamReaderInterface
QXmlStreamReader	m_qxmlStreamReader

Detailed Description

Todo: write docs

Definition at line 50 of file mzidentmlreader.h.

Member Enumeration Documentation

◆ IdentificationEngine

enum class pappso::cbor::psm::MzIdentMlReader::IdentificationEngine : std::int8_t

strongprivate

Enumerator
unknown	X!Tandem.
XTandem	MS:1001476 X!Tandem was used to analyze the spectra.
mascot	MS:1001207 The name of the Mascot search engine.
peptider	peptider
OMSSA	MS:1001475 Open Mass Spectrometry Search Algorithm was used to analyze the spectra.
SEQUEST	MS:1001208 The name of the SEQUEST search engine.
Comet	MS:1002251 Comet open-source sequence search engine developed at the University of Washington. PMID:23148064
Morpheus	MS:1002661 "Morpheus search engine." [PMID:23323968].
MSGFplus	MS:1002048 "MS-GF+ software used to analyze the spectra." [PSI:PI].
SpecOMS	SpecOMS C++ implementation.
sage	sage
PEAKS_Studio	PEAKS Studio.

Definition at line 91 of file mzidentmlreader.h.

  {
    unknown  = 0,      ///< X!Tandem
    XTandem  = 1,      ///< MS:1001476 X!Tandem was used to analyze the spectra.
    mascot   = 2,      ///< MS:1001207 The name of the Mascot search engine.
    peptider = 3,      ///< peptider
    OMSSA    = 4,      ///< MS:1001475 Open Mass Spectrometry Search Algorithm was used to
                       ///< analyze the spectra.
    SEQUEST = 5,       ///< MS:1001208 The name of the SEQUEST search engine.
    Comet   = 6,       ///< MS:1002251 Comet open-source sequence search engine developed
                       ///< at the University of Washington. PMID:23148064
    Morpheus     = 7,  ///< MS:1002661 "Morpheus search engine." [PMID:23323968]
    MSGFplus     = 8,  ///< MS:1002048 "MS-GF+ software used to analyze the spectra." [PSI:PI]
    SpecOMS      = 9,  ///< SpecOMS C++ implementation
    sage         = 10, ///< sage
    PEAKS_Studio = 11, ///< PEAKS Studio
  };

Constructor & Destructor Documentation

◆ MzIdentMlReader()

pappso::cbor::psm::MzIdentMlReader::MzIdentMlReader	(	pappso::UiMonitorInterface *	p_monitor,
		pappso::cbor::CborStreamWriter *	p_output,
		const QFileInfo &	mzident_file )

Default constructor

Definition at line 62 of file mzidentmlreader.cpp.

{
 
  mp_monitor = p_monitor;
  qDebug() << mzident_file.absoluteFilePath() << "'";
  m_identificationEngine = IdentificationEngine::unknown;
  m_mzidentFile          = mzident_file;
 
 
  try
    {
 
      mp_cborWriter = p_output;
 
      mp_cborWriter->startMap();
      mp_cborWriter->writeInformations(
        "mzidentml_file_reader", Utils::getVersion(), "psm", "mzidentml reader");
 
 
      if(!readFile(mzident_file.absoluteFilePath()))
        {
 
          if(errorString() == "Not an MzIdentML file")
            {
              throw pappso::ExceptionNotRecognized(
                QObject::tr("Error reading %1 not mzIdentML file :\n %2")
                  .arg(mzident_file.absoluteFilePath())
                  .arg(errorString()));
            }
          else
            {
              throw pappso::PappsoException(QObject::tr("Error reading %1 mzIdentML file :\n %2")
                                              .arg(mzident_file.absoluteFilePath())
                                              .arg(errorString()));
            }
        }
 
 
      mp_cborWriter->endMap();
    }
  catch(pappso::ExceptionNotRecognized &err)
    {
      throw err;
    }
  catch(pappso::PappsoException &other_err)
    {
      throw pappso::PappsoException(QObject::tr("Error reading mzIdentML file %1:\n%2")
                                      .arg(mzident_file.absoluteFilePath())
                                      .arg(other_err.qwhat()));
    }
}

References pappso::XmlStreamReaderInterface::errorString(), pappso::Utils::getVersion(), m_identificationEngine, m_mzidentFile, mp_cborWriter, mp_monitor, pappso::PappsoException::qwhat(), pappso::XmlStreamReaderInterface::readFile(), and unknown.

◆ ~MzIdentMlReader()

pappso::cbor::psm::MzIdentMlReader::~MzIdentMlReader ( )

virtual

Destructor

Definition at line 116 of file mzidentmlreader.cpp.

117{

118}

Member Function Documentation

◆ finalDebrief()

void pappso::cbor::psm::MzIdentMlReader::finalDebrief ( )

private

Definition at line 122 of file mzidentmlreader.cpp.

{
  qDebug();
  mp_cborWriter->append("parameter_map");
  mp_cborWriter->writeCborMap(m_cborParameterMap);
 
  QStringList fasta_files;
  for(auto &pair_searchdb : m_mzidSearchDatabaseIdMap)
    {
      fasta_files << pair_searchdb.second.file;
    }
  mp_cborWriter->append("target_fasta_files");
  mp_cborWriter->writeArray(fasta_files);
 
 
  for(std::pair<const QString, MzidDBSequence> &pair_sequence : m_MzidDBSequenceIdMap)
    {
      PsmProtein psm_protein;
 
      psm_protein.isContaminant = false;
      psm_protein.isTarget      = !pair_sequence.second.is_decoy;
      psm_protein.protein_sp    = pair_sequence.second.protein_sp;
 
      m_proteinMap.insert(psm_protein);
    }
 
  mp_cborWriter->append(QString("protein_map"));
  m_proteinMap.writeMap(*mp_cborWriter);
 
  mp_cborWriter->append("sample_list");
  mp_cborWriter->startArray();
  for(std::pair<const QString, std::vector<SpectrumIdentificationResult>> &pair_spectra :
      m_spectrumIdentificationResultBySpectraIdMap)
    {
 
      mp_cborWriter->startMap();
      mp_cborWriter->append("name");
      mp_cborWriter->append(m_mzidSpectraDataIdMap.at(pair_spectra.first).name);
 
      mp_cborWriter->append("identification_file_list");
      mp_cborWriter->startMap();
      mp_cborWriter->append("name");
      mp_cborWriter->append(m_mzidentFile.absoluteFilePath());
      mp_cborWriter->endMap();
      //  one_sample.cbor_core_sample.value("identification_file_list")
      //  .toCbor(m_sageReader.getCborStreamWriter());
 
 
      mp_cborWriter->append("peaklist_file");
      mp_cborWriter->startMap();
      mp_cborWriter->append("name");
      mp_cborWriter->append(m_mzidSpectraDataIdMap.at(pair_spectra.first).file);
      mp_cborWriter->endMap();
      //"scan_list": [
 
      mp_cborWriter->append("scan_list");
      mp_cborWriter->startArray(pair_spectra.second.size());
      for(SpectrumIdentificationResult &it_spectrum_ident : pair_spectra.second)
        {
          writeSpectrumIdentificationResult(it_spectrum_ident);
        }
      mp_cborWriter->endArray();
 
 
      mp_cborWriter->endMap();
    }
 
  mp_cborWriter->endArray();
  qDebug();
}

References pappso::cbor::psm::PsmProtein::isContaminant, pappso::cbor::psm::PsmProtein::isTarget, m_cborParameterMap, m_MzidDBSequenceIdMap, m_mzidentFile, m_mzidSearchDatabaseIdMap, m_mzidSpectraDataIdMap, m_proteinMap, m_spectrumIdentificationResultBySpectraIdMap, mp_cborWriter, pappso::cbor::psm::PsmProtein::protein_sp, and writeSpectrumIdentificationResult().

Referenced by readAnalysisData().

◆ readAnalysisCollectionItem()

bool pappso::cbor::psm::MzIdentMlReader::readAnalysisCollectionItem ( )

private

Definition at line 817 of file mzidentmlreader.cpp.

{
 
  qDebug();
  if(m_qxmlStreamReader.readNextStartElement())
    {
      if(m_qxmlStreamReader.name().toString() == "SpectrumIdentification")
        {
          readSpectrumIdentification();
          qDebug();
        }
      else if(m_qxmlStreamReader.name().toString() == "ProteinDetection")
        {
          m_qxmlStreamReader.skipCurrentElement();
        }
      else
        {
          m_qxmlStreamReader.raiseError(
            QObject::tr("Error in MzIdentML/AnalysisCollection, unexpected %1 t")
              .arg(m_qxmlStreamReader.name()));
          m_qxmlStreamReader.skipCurrentElement();
          return false;
        }
      return true;
    }
  return false;
}

References pappso::XmlStreamReaderInterface::m_qxmlStreamReader, and readSpectrumIdentification().

Referenced by readStream().

◆ readAnalysisData()

void pappso::cbor::psm::MzIdentMlReader::readAnalysisData ( )

private

Definition at line 981 of file mzidentmlreader.cpp.

{
 
  qDebug() << m_qxmlStreamReader.name();
  while(m_qxmlStreamReader.readNextStartElement())
    {
      qDebug() << m_qxmlStreamReader.name();
      if(m_qxmlStreamReader.name().toString() == "SpectrumIdentificationList")
        {
          while(m_qxmlStreamReader.readNextStartElement())
            {
              qDebug() << m_qxmlStreamReader.name();
              if(m_qxmlStreamReader.name().toString() == "SpectrumIdentificationResult")
                {
                  readSpectrumIdentificationResult();
                }
              else if(m_qxmlStreamReader.name().toString() == "FragmentationTable")
                {
                  m_qxmlStreamReader.skipCurrentElement();
                }
              else
                {
                  m_qxmlStreamReader.raiseError(
                    QObject::tr("Error in "
                                "MzIdentML/DataCollection/AnalysisData/"
                                "SpectrumIdentificationList, unexpected %1 tag")
                      .arg(m_qxmlStreamReader.name()));
                  m_qxmlStreamReader.skipCurrentElement();
                }
            }
        }
      else if(m_qxmlStreamReader.name().toString() == "ProteinDetectionList")
        {
          // ProteinDetectionList
          m_qxmlStreamReader.skipCurrentElement();
        }
      else
        {
          m_qxmlStreamReader.raiseError(
            QObject::tr("Error in MzIdentML/DataCollection/AnalysisData, "
                        "unexpected %1 tag")
              .arg(m_qxmlStreamReader.name()));
          m_qxmlStreamReader.skipCurrentElement();
        }
    }
  qDebug();
  finalDebrief();
}

References finalDebrief(), pappso::XmlStreamReaderInterface::m_qxmlStreamReader, and readSpectrumIdentificationResult().

Referenced by readDataCollectionItem().

◆ readAnalysisProtocolCollection()

void pappso::cbor::psm::MzIdentMlReader::readAnalysisProtocolCollection ( )

private

Definition at line 276 of file mzidentmlreader.cpp.

{
  //<AnalysisProtocolCollection>
 
  qDebug();
  while(m_qxmlStreamReader.readNextStartElement())
    {
      qDebug() << m_qxmlStreamReader.name();
      if(m_qxmlStreamReader.name().toString() == "SpectrumIdentificationProtocol")
        {
          auto it_soft = m_IdentificationEngineMap.find(
            m_qxmlStreamReader.attributes().value("analysisSoftware_ref").toString());
          if(it_soft != m_IdentificationEngineMap.end())
            {
              IdentificationEngine identification_engine = it_soft->second;
 
              if(identification_engine == IdentificationEngine::XTandem)
                {
                  //<SpectrumIdentificationProtocol analysisSoftware_ref="ID_software"
                  // id="SearchProtocol_1">
                  QCborMap identification_engine_parameters;
 
                  while(m_qxmlStreamReader.readNextStartElement())
                    {
                      if(m_qxmlStreamReader.name().toString() == "SearchType")
                        {
                          m_qxmlStreamReader.skipCurrentElement();
                        }
                      else if(m_qxmlStreamReader.name().toString() == "ModificationParams")
                        {
                          m_qxmlStreamReader.skipCurrentElement();
                        }
 
                      else if(m_qxmlStreamReader.name().toString() == "Threshold")
                        {
                          m_qxmlStreamReader.skipCurrentElement();
                        }
                      else if(m_qxmlStreamReader.name().toString() == "ParentTolerance")
                        {
                          m_qxmlStreamReader.skipCurrentElement();
                        }
                      else if(m_qxmlStreamReader.name().toString() == "FragmentTolerance")
                        {
                          m_qxmlStreamReader.skipCurrentElement();
                        }
                      else if(m_qxmlStreamReader.name().toString() == "Enzymes")
                        {
                          m_qxmlStreamReader.skipCurrentElement();
                        }
                      else if(m_qxmlStreamReader.name().toString() == "AdditionalSearchParams")
                        {
                          while(m_qxmlStreamReader.readNextStartElement())
                            {
                              if(m_qxmlStreamReader.name().toString() == "userParam")
                                {
                                  //<userParam name="list path, default parameters"
                                  // value="/tmp/i2masschroq.ubFSuT/QExactive_analysis_FDR_nosemi.xml"/>
                                  identification_engine_parameters.insert(
                                    m_qxmlStreamReader.attributes().value("name").toString(),
                                    m_qxmlStreamReader.attributes().value("value").toString());
                                  m_qxmlStreamReader.skipCurrentElement();
                                }
                              else
                                {
                                  m_qxmlStreamReader.skipCurrentElement();
                                }
                            }
                        }
                      else
                        {
 
                          m_qxmlStreamReader.raiseError(QObject::tr("element %1 not implemented")
                                                          .arg(m_qxmlStreamReader.name()));
                          m_qxmlStreamReader.skipCurrentElement();
                        }
                    }
 
                  m_cborParameterMap.insert(QString("xtandem"),
                                            identification_engine_parameters.toCborValue());
                  //  qWarning() << "coucou";
                }
              else
                {
                  // other than tandem
                  m_qxmlStreamReader.skipCurrentElement();
                }
            }
          else
            {
              m_qxmlStreamReader.skipCurrentElement();
            }
        }
      else
        {
          m_qxmlStreamReader.raiseError(
            QObject::tr("element %1 not implemented").arg(m_qxmlStreamReader.name()));
          m_qxmlStreamReader.skipCurrentElement();
        }
    }
  qDebug();
}

References m_cborParameterMap, m_IdentificationEngineMap, pappso::XmlStreamReaderInterface::m_qxmlStreamReader, and XTandem.

Referenced by readStream().

◆ readAnalysisSoftware()

bool pappso::cbor::psm::MzIdentMlReader::readAnalysisSoftware ( )

private

stores the current analysis software id

tells if the software name has been found and is handled by the parser

Definition at line 379 of file mzidentmlreader.cpp.

{
 
 
  /** @brief stores the current analysis software id
   */
  QString analysisSoftwareId;
 
 
  /** @brief tells if the software name has been found and is handled by the
   * parser
   */
  IdentificationEngine analysisSotwareNameFound = IdentificationEngine::unknown;
  if(m_qxmlStreamReader.readNextStartElement())
    {
      if(m_qxmlStreamReader.name().toString() == "AnalysisSoftware")
        {
 
          // <AnalysisSoftware version="0.0.9" name="DeepProt" id="as1">
 
 
          analysisSoftwareId    = m_qxmlStreamReader.attributes().value("id").toString();
          QString software_name = m_qxmlStreamReader.attributes().value("name").toString();
 
 
          m_analysisSoftwareVersion = m_qxmlStreamReader.attributes().value("version").toString();
 
          if(software_name == "SpecOMS")
            {
              analysisSotwareNameFound = IdentificationEngine::SpecOMS;
 
              m_qxmlStreamReader.skipCurrentElement();
            }
          else if(software_name == "DeepProt")
            {
              analysisSotwareNameFound = IdentificationEngine::SpecOMS;
 
              m_qxmlStreamReader.skipCurrentElement();
            }
          else
            {
              while(m_qxmlStreamReader.readNextStartElement())
                {
                  if(m_qxmlStreamReader.name().toString() == "SoftwareName")
                    {
                      while(m_qxmlStreamReader.readNextStartElement())
                        {
                          if(m_qxmlStreamReader.name().toString() == "cvParam")
                            {
                              CvParam cv_param = readCvParam();
                              if(cv_param.accession == "MS:1001476")
                                {
                                  analysisSotwareNameFound = IdentificationEngine::XTandem;
                                }
                              else if(cv_param.accession == "MS:1002048")
                                {
                                  analysisSotwareNameFound = IdentificationEngine::MSGFplus;
                                }
                              else if(cv_param.accession == "MS:1001946")
                                {
                                  analysisSotwareNameFound = IdentificationEngine::PEAKS_Studio;
                                }
                            }
                        }
                    }
                  else
                    {
                      m_qxmlStreamReader.skipCurrentElement();
                    }
                }
            }
 
 
          switch(analysisSotwareNameFound)
            {
              case IdentificationEngine::SpecOMS:
                break;
              case IdentificationEngine::MSGFplus:
                break;
              case IdentificationEngine::XTandem:
                break;
              case IdentificationEngine::PEAKS_Studio:
                break;
 
              default:
                m_qxmlStreamReader.raiseError(
                  QObject::tr("identification results from %1 are not supported yet, "
                              "Please contact "
                              "the PAPPSO team.")
                    .arg(software_name));
                return false;
            }
 
 
          auto it = m_IdentificationEngineMap.insert(
            std::pair<QString, IdentificationEngine>(analysisSoftwareId, analysisSotwareNameFound));
 
          if(it.second == false)
            {
              it.first->second = analysisSotwareNameFound;
            }
        }
      else
        {
          m_qxmlStreamReader.raiseError(QObject::tr("Not an MzIdentML input file"));
          m_qxmlStreamReader.skipCurrentElement();
          return false;
        }
      return true;
    }
  return false;
}

References pappso::cbor::psm::MzIdentMlReader::CvParam::accession, m_analysisSoftwareVersion, m_IdentificationEngineMap, pappso::XmlStreamReaderInterface::m_qxmlStreamReader, MSGFplus, PEAKS_Studio, readCvParam(), SpecOMS, unknown, and XTandem.

Referenced by readStream().

◆ readCvParam()

MzIdentMlReader::CvParam pappso::cbor::psm::MzIdentMlReader::readCvParam ( )

private

Definition at line 592 of file mzidentmlreader.cpp.

{
  CvParam cv_param;
  cv_param.cvRef         = m_qxmlStreamReader.attributes().value("cvRef").toString();
  cv_param.accession     = m_qxmlStreamReader.attributes().value("accession").toString();
  cv_param.name          = m_qxmlStreamReader.attributes().value("name").toString();
  cv_param.value         = m_qxmlStreamReader.attributes().value("value").toString();
  cv_param.unitAccession = m_qxmlStreamReader.attributes().value("unitAccession").toString();
  cv_param.unitName      = m_qxmlStreamReader.attributes().value("unitName").toString();
  cv_param.unitCvRef     = m_qxmlStreamReader.attributes().value("unitCvRef").toString();
  m_qxmlStreamReader.skipCurrentElement();
  return cv_param;
}

References pappso::cbor::psm::MzIdentMlReader::CvParam::accession, pappso::cbor::psm::MzIdentMlReader::CvParam::cvRef, pappso::XmlStreamReaderInterface::m_qxmlStreamReader, pappso::cbor::psm::MzIdentMlReader::CvParam::name, pappso::cbor::psm::MzIdentMlReader::CvParam::unitAccession, pappso::cbor::psm::MzIdentMlReader::CvParam::unitCvRef, pappso::cbor::psm::MzIdentMlReader::CvParam::unitName, and pappso::cbor::psm::MzIdentMlReader::CvParam::value.

Referenced by readAnalysisSoftware(), readDBSequence(), readPeptide(), readSpectrumIdentificationItem(), and readSpectrumIdentificationResult().

◆ readDataCollectionItem()

bool pappso::cbor::psm::MzIdentMlReader::readDataCollectionItem ( )

private

Definition at line 853 of file mzidentmlreader.cpp.

{
 
  qDebug();
  if(m_qxmlStreamReader.readNextStartElement())
    {
      if(m_qxmlStreamReader.name().toString() == "Inputs")
        {
          readInputs();
        }
 
      // AnalysisData
      else if(m_qxmlStreamReader.name().toString() == "AnalysisData")
        {
          qDebug();
          readAnalysisData();
        }
      else
        {
          m_qxmlStreamReader.raiseError(
            QObject::tr("Error in MzIdentML/DataCollection, unexpected %1 tag")
              .arg(m_qxmlStreamReader.name()));
          m_qxmlStreamReader.skipCurrentElement();
          return false;
        }
      return true;
    }
  return false;
}

References pappso::XmlStreamReaderInterface::m_qxmlStreamReader, readAnalysisData(), and readInputs().

Referenced by readStream().

◆ readDBSequence()

void pappso::cbor::psm::MzIdentMlReader::readDBSequence ( )

private

Definition at line 524 of file mzidentmlreader.cpp.

{
 
  // attributes.value("base_name")
  // ProteinXtpSp sp_xtp_protein = _current_protein.makeProteinXtpSp();
  MzidDBSequence dbsequence;
  dbsequence.is_decoy  = false;
  dbsequence.accession = m_qxmlStreamReader.attributes().value("accession").toString();
 
  QString accession_description = dbsequence.accession;
  dbsequence.searchDatabase_ref =
    m_qxmlStreamReader.attributes().value("searchDatabase_ref").toString();
 
  QString id = m_qxmlStreamReader.attributes().value("id").toString();
 
 
  while(m_qxmlStreamReader.readNextStartElement())
    {
      if(m_qxmlStreamReader.name().toString() == "cvParam")
        {
          CvParam cv_param = readCvParam();
 
 
          if(cv_param.accession == "MS:1001088")
            {
              // protein description
              // protein.get()->setDescription(cv_param.value);
              dbsequence.description = cv_param.value;
              accession_description.append(" ").append(dbsequence.description);
            }
          else if(cv_param.accession == "MS:1001195")
            {
              // PSI-MS MS:1001195 decoy DB type reverse
              // protein.get()->setIsDecoy(true);
              dbsequence.is_decoy = true;
            }
          else
            {
              dbsequence.cvParamList.push_back(cv_param);
            }
        }
      else if(m_qxmlStreamReader.name().toString().toLower() == "seq")
        {
 
 
          dbsequence.sequence = m_qxmlStreamReader.readElementText();
        }
      else
        {
 
          m_qxmlStreamReader.raiseError(
            QObject::tr("Error in MzIdentML/DBSequence unexpected %1 tag")
              .arg(m_qxmlStreamReader.name()));
          m_qxmlStreamReader.skipCurrentElement();
        }
    }
 
 
  // qWarning() << accession_description;
  dbsequence.protein_sp = std::make_shared<Protein>(accession_description, dbsequence.sequence);
 
  m_MzidDBSequenceIdMap.insert({id, dbsequence});
  // searchDatabase_ref="SearchDB_1"
 
  // protein.get()->setFastaFileP(fastaFile.get());
}

References pappso::cbor::psm::MzIdentMlReader::CvParam::accession, pappso::cbor::psm::MzIdentMlReader::MzidDBSequence::accession, pappso::cbor::psm::MzIdentMlReader::MzidDBSequence::cvParamList, pappso::cbor::psm::MzIdentMlReader::MzidDBSequence::description, pappso::cbor::psm::MzIdentMlReader::MzidDBSequence::is_decoy, m_MzidDBSequenceIdMap, pappso::XmlStreamReaderInterface::m_qxmlStreamReader, pappso::cbor::psm::MzIdentMlReader::MzidDBSequence::protein_sp, readCvParam(), pappso::cbor::psm::MzIdentMlReader::MzidDBSequence::searchDatabase_ref, pappso::cbor::psm::MzIdentMlReader::MzidDBSequence::sequence, and pappso::cbor::psm::MzIdentMlReader::CvParam::value.

Referenced by readSequenceCollectionItem().

◆ readInputs()

void pappso::cbor::psm::MzIdentMlReader::readInputs ( )

private

Definition at line 884 of file mzidentmlreader.cpp.

{
  while(m_qxmlStreamReader.readNextStartElement())
    {
      if(m_qxmlStreamReader.name().toString() == "SearchDatabase")
        {
          readSearchDatabase();
        }
      else if(m_qxmlStreamReader.name().toString() == "SpectraData")
        {
          readSpectraData();
        }
      else if(m_qxmlStreamReader.name().toString() == "SourceFile")
        {
          m_qxmlStreamReader.skipCurrentElement();
        }
 
      else
        {
          m_qxmlStreamReader.raiseError(
            QObject::tr("Error in MzIdentML/DataCollection/Inputs, unexpected %1 tag")
              .arg(m_qxmlStreamReader.name()));
          m_qxmlStreamReader.skipCurrentElement();
        }
    }
}

References pappso::XmlStreamReaderInterface::m_qxmlStreamReader, readSearchDatabase(), and readSpectraData().

Referenced by readDataCollectionItem().

◆ readPeptide()

void pappso::cbor::psm::MzIdentMlReader::readPeptide ( )

private

Definition at line 620 of file mzidentmlreader.cpp.

{
  std::shared_ptr<pappso::Peptide> peptide_sp;
 
  QString xml_id = m_qxmlStreamReader.attributes().value("id").toString();
 
  // PeptideSequence
  if(m_qxmlStreamReader.readNextStartElement())
    {
      if(m_qxmlStreamReader.name().toString() == "PeptideSequence")
        {
          peptide_sp = std::make_shared<Peptide>(m_qxmlStreamReader.readElementText().simplified());
        }
      else
        {
 
          m_qxmlStreamReader.raiseError(
            QObject::tr("Error in MzIdentML/Peptide no PeptideSequence"));
        }
    }
 
  while(m_qxmlStreamReader.readNextStartElement())
    {
      if(m_qxmlStreamReader.name().toString() == "Modification")
        {
          //      <Modification monoisotopicMassDelta="-0.02682025649"
          //      location="1">
          //          <cvParam accession="-0.0268203" cvRef="PSI-MOD" name=""/>
          //      </Modification>
 
          Modification modification;
          modification.monoisotopicMassDelta =
            m_qxmlStreamReader.attributes().value("monoisotopicMassDelta").toDouble();
          modification.location = m_qxmlStreamReader.attributes().value("location").toUInt();
          bool is_cv_param      = false;
          while(m_qxmlStreamReader.readNextStartElement())
            {
              if(m_qxmlStreamReader.name().toString() == "cvParam")
                {
                  is_cv_param          = true;
                  modification.cvParam = readCvParam();
 
 
                  // qDebug() << "startElement_aa ";
                  pappso::AaModificationP modif = nullptr;
 
 
                  qDebug() << modification.monoisotopicMassDelta;
                  if(modification.cvParam.accession != "")
                    {
                      qDebug() << modification.cvParam.accession;
                      if(modification.cvParam.accession.startsWith("UNIMOD:"))
                        {
                          modif = pappso::Utils::translateAaModificationFromUnimod(
                            modification.cvParam.accession);
                        }
                      else
                        {
                          // hope it is psi mod:
                          if(modification.cvParam.accession.startsWith("MOD:"))
                            {
                              modif =
                                pappso::AaModification::getInstance(modification.cvParam.accession);
                            }
                          else if(modification.cvParam.accession == "MS:1001460")
                            {
                              //[Term]
                              // id: MS:1001460
                              // name: unknown modification
                              // def: "This term should be given if the
                              // modification was unknown." [PSI:PI] is_a:
                              // MS:1001471 ! peptide modification details
                              modif = pappso::AaModification::getInstanceCustomizedMod(
                                modification.monoisotopicMassDelta);
                            }
                          else
                            {
                              qInfo() << "MzIdentMlSaxHandler::endElement_"
                                         "Modification unknown "
                                         "modification "
                                      << modification.cvParam.accession << " "
                                      << modification.cvParam.name;
                            }
                        }
                    }
 
                  if(modif == nullptr)
                    {
                      m_qxmlStreamReader.raiseError(
                        QObject::tr("Error in MzIdentML/Peptide/Modification/cvParam "
                                    "modification accession %1 not found")
                          .arg(modification.cvParam.accession));
                      return;
                    }
 
                  if(modification.location == 0)
                    {
                      peptide_sp.get()->addAaModification(modif, 0);
                    }
                  else
                    {
                      peptide_sp.get()->addAaModification(modif, modification.location - 1);
                    }
                }
              else
                {
 
                  m_qxmlStreamReader.raiseError(
                    QObject::tr("Error in MzIdentML/Peptide/Modification "
                                "unexpected %1 tag")
                      .arg(m_qxmlStreamReader.name()));
                }
            }
 
          if(is_cv_param == false)
            {
              pappso::AaModificationP modif_without_cvparam = nullptr;
              // no cv param element
 
              if(modification.location == 0)
                {
                  modif_without_cvparam =
                    pappso::Utils::guessAaModificationPbyMonoisotopicMassDelta(
                      peptide_sp.get()->getAa(0).getAminoAcidChar(),
                      modification.monoisotopicMassDelta);
                  peptide_sp.get()->addAaModification(modif_without_cvparam, 0);
                }
              else
                {
                  modif_without_cvparam =
                    pappso::Utils::guessAaModificationPbyMonoisotopicMassDelta(
                      peptide_sp.get()->getAa(modification.location - 1).getAminoAcidChar(),
                      modification.monoisotopicMassDelta);
                  peptide_sp.get()->addAaModification(modif_without_cvparam,
                                                      modification.location - 1);
                }
            }
        }
 
      else
        {
 
          m_qxmlStreamReader.raiseError(QObject::tr("Error in MzIdentML/Peptide unexpected %1 tag")
                                          .arg(m_qxmlStreamReader.name()));
        }
    }
 
 
  m_PeptideIdMap.insert(std::pair<QString, PeptideSp>(xml_id, peptide_sp));
}

References pappso::cbor::psm::MzIdentMlReader::CvParam::accession, pappso::cbor::psm::MzIdentMlReader::Modification::cvParam, pappso::AaModification::getInstance(), pappso::AaModification::getInstanceCustomizedMod(), pappso::Utils::guessAaModificationPbyMonoisotopicMassDelta(), pappso::cbor::psm::MzIdentMlReader::Modification::location, m_PeptideIdMap, pappso::XmlStreamReaderInterface::m_qxmlStreamReader, pappso::cbor::psm::MzIdentMlReader::Modification::monoisotopicMassDelta, pappso::cbor::psm::MzIdentMlReader::CvParam::name, readCvParam(), and pappso::Utils::translateAaModificationFromUnimod().

Referenced by readSequenceCollectionItem().

◆ readPeptideEvidence()

bool pappso::cbor::psm::MzIdentMlReader::readPeptideEvidence ( )

private

Definition at line 772 of file mzidentmlreader.cpp.

{
  qDebug();
  MzidPeptideEvidence pe;
 
  auto itprot =
    m_MzidDBSequenceIdMap.find(m_qxmlStreamReader.attributes().value("dBSequence_ref").toString());
  if(itprot == m_MzidDBSequenceIdMap.end())
    {
      m_qxmlStreamReader.raiseError(
        QObject::tr("dBSequence_ref %1 not defined")
          .arg(m_qxmlStreamReader.attributes().value("dBSequence_ref").toString()));
      return false;
    }
  qDebug();
  pe.protein = itprot->second.protein_sp;
 
 
  auto itpep = m_PeptideIdMap.find(m_qxmlStreamReader.attributes().value("peptide_ref").toString());
  if(itpep == m_PeptideIdMap.end())
    {
      m_qxmlStreamReader.raiseError(
        QObject::tr("peptide_ref %1 not defined")
          .arg(m_qxmlStreamReader.attributes().value("peptide_ref").toString()));
      return false;
    }
  qDebug();
  pe.peptide = itpep->second;
 
  pe.start   = m_qxmlStreamReader.attributes().value("start").toUInt() - 1;
  pe.end     = m_qxmlStreamReader.attributes().value("end").toUInt() - 1;
  pe.isDecoy = false;
  if(m_qxmlStreamReader.attributes().value("isDecoy").toString() == "true")
    {
      pe.isDecoy = true;
    }
  qDebug();
  m_MzidPeptideEvidenceIdMap.insert(std::pair<QString, MzidPeptideEvidence>(
    m_qxmlStreamReader.attributes().value("id").toString(), pe));
  m_qxmlStreamReader.skipCurrentElement();
  qDebug();
  return true;
}

References pappso::cbor::psm::MzIdentMlReader::MzidPeptideEvidence::end, pappso::cbor::psm::MzIdentMlReader::MzidPeptideEvidence::isDecoy, m_MzidDBSequenceIdMap, m_MzidPeptideEvidenceIdMap, m_PeptideIdMap, pappso::XmlStreamReaderInterface::m_qxmlStreamReader, pappso::cbor::psm::MzIdentMlReader::MzidPeptideEvidence::peptide, pappso::cbor::psm::MzIdentMlReader::MzidPeptideEvidence::protein, and pappso::cbor::psm::MzIdentMlReader::MzidPeptideEvidence::start.

Referenced by readSequenceCollectionItem().

◆ readSearchDatabase()

bool pappso::cbor::psm::MzIdentMlReader::readSearchDatabase ( )

private

Definition at line 923 of file mzidentmlreader.cpp.

{
  /* <SearchDatabase
     location="/gorgone/pappso/versions_logiciels_pappso/tandemng/database/Genome_Z_mays_5a.fasta"
     id="SearchDB_0"> <FileFormat> <cvParam accession="MS:1001348" cvRef="PSI-MS" name="FASTA
     format"/>
                  </FileFormat>
                  <DatabaseName>
                      <userParam name="DatabaseName" value="Genome_Z_mays_5a.fasta"/>
                  </DatabaseName>
                  <cvParam accession="MS:1001197" cvRef="PSI-MS" name="DB composition
     target+decoy"/> <cvParam accession="MS: 1001283" cvRef="PSI-MS" name="decoy DB accession
     regexp" value="^XXX"/> <cvParam accession="MS: 1001195" cvRef="PSI-MS" name="decoy DB type
     reverse"/>
              </SearchDatabase>
              */
  qDebug();
  QString id = m_qxmlStreamReader.attributes().value("id").toString();
  // auto itfasta = m_FastaFileIdMap.find(id);
 
  MzidSearchDatabase search_database;
 
  search_database.file = m_qxmlStreamReader.attributes().value("location").toString();
 
 
  if(search_database.file.isEmpty())
    {
      m_qxmlStreamReader.raiseError(QObject::tr("SearchDatabase id %1 location is empty").arg(id));
      return false;
    }
 
 
  m_mzidSearchDatabaseIdMap.insert({id, search_database});
 
  m_qxmlStreamReader.skipCurrentElement();
  qDebug();
  return true;
}

References pappso::cbor::psm::MzIdentMlReader::MzidSearchDatabase::file, m_mzidSearchDatabaseIdMap, and pappso::XmlStreamReaderInterface::m_qxmlStreamReader.

Referenced by readInputs().

◆ readSequenceCollectionItem()

bool pappso::cbor::psm::MzIdentMlReader::readSequenceCollectionItem ( )

private

Definition at line 493 of file mzidentmlreader.cpp.

{
  if(m_qxmlStreamReader.readNextStartElement())
    {
      if(m_qxmlStreamReader.name().toString() == "DBSequence")
        {
          readDBSequence();
        }
      else if(m_qxmlStreamReader.name().toString() == "Peptide")
        {
          readPeptide();
        }
      // PeptideEvidence
      else if(m_qxmlStreamReader.name().toString() == "PeptideEvidence")
        {
          readPeptideEvidence();
        }
      else
        {
          m_qxmlStreamReader.raiseError(
            QObject::tr("Error in MzIdentML input file, %1 no DBSequence")
              .arg(m_qxmlStreamReader.name()));
          m_qxmlStreamReader.skipCurrentElement();
          return false;
        }
      return true;
    }
  return false;
}

References pappso::XmlStreamReaderInterface::m_qxmlStreamReader, readDBSequence(), readPeptide(), and readPeptideEvidence().

Referenced by readStream().

◆ readSpectraData()

void pappso::cbor::psm::MzIdentMlReader::readSpectraData ( )

private

Definition at line 963 of file mzidentmlreader.cpp.

{
 
  qDebug();
  MzidSpectraData spectra_data;
  spectra_data.file = m_qxmlStreamReader.attributes().value("location").toString();
 
 
  // msrun.get()->setXmlId(attributes.value("id"));
  spectra_data.name = m_qxmlStreamReader.attributes().value("name").toString();
 
 
  m_mzidSpectraDataIdMap.insert(std::pair<QString, MzidSpectraData>(
    m_qxmlStreamReader.attributes().value("id").toString(), spectra_data));
  m_qxmlStreamReader.skipCurrentElement();
}

References pappso::cbor::psm::MzIdentMlReader::MzidSpectraData::file, m_mzidSpectraDataIdMap, pappso::XmlStreamReaderInterface::m_qxmlStreamReader, and pappso::cbor::psm::MzIdentMlReader::MzidSpectraData::name.

Referenced by readInputs().

◆ readSpectrumIdentification()

void pappso::cbor::psm::MzIdentMlReader::readSpectrumIdentification ( )

private

Definition at line 846 of file mzidentmlreader.cpp.

{
 
  m_qxmlStreamReader.skipCurrentElement();
}

References pappso::XmlStreamReaderInterface::m_qxmlStreamReader.

Referenced by readAnalysisCollectionItem().

◆ readSpectrumIdentificationItem()

void pappso::cbor::psm::MzIdentMlReader::readSpectrumIdentificationItem ( MzIdentMlReader::SpectrumIdentificationResult & spectrum_identification_result )

private

Definition at line 1179 of file mzidentmlreader.cpp.

{
 
  qDebug();
  spectrum_identification_result.spectrumIdentificationItemList.push_back(
    SpectrumIdentificationItem());
  spectrum_identification_result.spectrumIdentificationItemList.back()
    .mzidPeptideEvidenceList.clear();
  spectrum_identification_result.spectrumIdentificationItemList.back().cvParamList.clear();
  spectrum_identification_result.spectrumIdentificationItemList.back().userParamList.clear();
 
  spectrum_identification_result.spectrumIdentificationItemList.back().chargeState =
    m_qxmlStreamReader.attributes().value("chargeState").toUInt();
 
  spectrum_identification_result.spectrumIdentificationItemList.back().experimentalMassToCharge =
    m_qxmlStreamReader.attributes().value("experimentalMassToCharge").toDouble();
 
 
  auto itpeptide =
    m_PeptideIdMap.find(m_qxmlStreamReader.attributes().value("peptide_ref").toString());
 
  if(itpeptide == m_PeptideIdMap.end())
    {
      m_qxmlStreamReader.raiseError(QObject::tr("peptide_ref %1 not defined")
                                      .arg(m_qxmlStreamReader.attributes().value("peptide_ref")));
    }
  spectrum_identification_result.spectrumIdentificationItemList.back().peptide = itpeptide->second;
 
 
  while(m_qxmlStreamReader.readNextStartElement())
    {
      qDebug() << m_qxmlStreamReader.name();
      if(m_qxmlStreamReader.name() == QString("PeptideEvidenceRef"))
        {
 
          auto itpeptideEvidence = m_MzidPeptideEvidenceIdMap.find(
            m_qxmlStreamReader.attributes().value("peptideEvidence_ref").toString());
 
          if(itpeptideEvidence == m_MzidPeptideEvidenceIdMap.end())
            {
              m_qxmlStreamReader.raiseError(
                QObject::tr("peptideEvidence_ref %1 not defined")
                  .arg(m_qxmlStreamReader.attributes().value("peptideEvidence_ref")));
            }
 
          spectrum_identification_result.spectrumIdentificationItemList.back()
            .mzidPeptideEvidenceList.push_back(itpeptideEvidence->second);
          m_qxmlStreamReader.skipCurrentElement();
        }
      else if(m_qxmlStreamReader.name() == QString("cvParam"))
        {
          spectrum_identification_result.spectrumIdentificationItemList.back()
            .cvParamList.push_back(readCvParam());
          qDebug() << spectrum_identification_result.spectrumIdentificationItemList.back()
                        .cvParamList.back()
                        .toString();
        }
      else if(m_qxmlStreamReader.name() == QString("userParam"))
        {
          UserParam user_param = readUserParam();
          spectrum_identification_result.spectrumIdentificationItemList.back()
            .userParamList.push_back(user_param);
          qDebug() << user_param.toString();
        }
      else
        {
          m_qxmlStreamReader.raiseError(QObject::tr("Error in "
                                                    "MzIdentML/DataCollection/AnalysisData/"
                                                    "SpectrumIdentificationList/"
                                                    "SpectrumIdentificationResult/"
                                                    "SpectrumIdentificationItem, unexpected %1 tag")
                                          .arg(m_qxmlStreamReader.name()));
        }
    }
  qDebug();
}

References m_MzidPeptideEvidenceIdMap, m_PeptideIdMap, pappso::XmlStreamReaderInterface::m_qxmlStreamReader, readCvParam(), readUserParam(), pappso::cbor::psm::MzIdentMlReader::SpectrumIdentificationResult::spectrumIdentificationItemList, and pappso::cbor::psm::MzIdentMlReader::UserParam::toString().

Referenced by readSpectrumIdentificationResult().

◆ readSpectrumIdentificationResult()

void pappso::cbor::psm::MzIdentMlReader::readSpectrumIdentificationResult ( )

private

Definition at line 1031 of file mzidentmlreader.cpp.

{
  qDebug() << m_qxmlStreamReader.name();
 
  QString spectra_id   = m_qxmlStreamReader.attributes().value("spectraData_ref").toString();
  auto it_spectra_data = m_mzidSpectraDataIdMap.find(spectra_id);
 
  if(it_spectra_data == m_mzidSpectraDataIdMap.end())
    {
      m_qxmlStreamReader.raiseError(QObject::tr("spectraData_ref %1 not defined in "
                                                "m_mzidSpectraDataIdMap")
                                      .arg(spectra_id));
    }
 
  auto it_insert_ident = m_spectrumIdentificationResultBySpectraIdMap.insert({spectra_id, {}});
 
  it_insert_ident.first->second.push_back({});
  SpectrumIdentificationResult &spectrum_identification_result =
    it_insert_ident.first->second.back();
  spectrum_identification_result.cvParamList.clear();
  spectrum_identification_result.userParamList.clear();
  spectrum_identification_result.spectrumIdentificationItemList.clear();
 
  spectrum_identification_result.spectrumID =
    m_qxmlStreamReader.attributes().value("spectrumID").toString();
  spectrum_identification_result.id = m_qxmlStreamReader.attributes().value("id").toString();
 
 
  qDebug() << m_qxmlStreamReader.name();
 
  while(m_qxmlStreamReader.readNextStartElement())
    {
      qDebug() << m_qxmlStreamReader.name();
      if(m_qxmlStreamReader.name() == QString("SpectrumIdentificationItem"))
        {
          qDebug();
          readSpectrumIdentificationItem(spectrum_identification_result);
        }
      else if(m_qxmlStreamReader.name() == QString("cvParam"))
        {
          CvParam cv_param = readCvParam();
          qDebug() << cv_param.toString();
          spectrum_identification_result.cvParamList.push_back(cv_param);
        }
 
      else if(m_qxmlStreamReader.name() == QString("userParam"))
        {
          UserParam user_param = readUserParam();
          qDebug() << user_param.toString();
        }
      else
        {
          m_qxmlStreamReader.raiseError(
            QObject::tr("Error in "
                        "MzIdentML/DataCollection/AnalysisData/"
                        "SpectrumIdentificationList/"
                        "SpectrumIdentificationResult, unexpected %1 tag")
              .arg(m_qxmlStreamReader.name()));
          // m_qxmlStreamReader.skipCurrentElement();
        }
    }
 
  if(m_qxmlStreamReader.hasError())
    return;
  qDebug() << m_qxmlStreamReader.name();
 
  // find scan number
  spectrum_identification_result.spectrumIndex   = 0;
  spectrum_identification_result.isSpectrumIndex = false;
  spectrum_identification_result.scanNum         = 0;
  spectrum_identification_result.retentionTime   = 0;
 
  // spectrumID="index=194"
  if(spectrum_identification_result.spectrumID.startsWith("index="))
    {
      bool is_ok = false;
      spectrum_identification_result.spectrumIndex =
        spectrum_identification_result.spectrumID.mid(6).toULongLong(&is_ok);
      spectrum_identification_result.isSpectrumIndex = true;
      if(!is_ok)
        {
          m_qxmlStreamReader.raiseError(QObject::tr("reading spectrum index failed in %1"
                                                    "SpectrumIdentificationResult id %2")
                                          .arg(spectrum_identification_result.spectrumID)
                                          .arg(spectrum_identification_result.id));
          return;
        }
    }
 
  //  <cvParam cvRef="PSI-MS" accession="MS:1001115" name="scan number(s)"
  //        value="16079"/>
  for(auto cvParam : spectrum_identification_result.cvParamList)
    {
 
      qDebug() << cvParam.toString();
      if(cvParam.accession == "MS:1001115")
        {
          spectrum_identification_result.scanNum = cvParam.value.toUInt();
        }
      else if(cvParam.accession == "MS:1003062")
        {
          spectrum_identification_result.isSpectrumIndex = true;
          spectrum_identification_result.spectrumIndex   = cvParam.value.toUInt();
        }
      else if((cvParam.accession == "MS:1000016") || (cvParam.accession == "MS:1000894"))
        {
          //[Term]
          // id: MS:1000894
          // name: retention time
          // def: "A time interval from the start of chromatography when an
          // analyte exits a chromatographic column." [PSI:MS]
 
          //            [Term]
          // id: MS:1000016
          // name: scan start time
          // def: "The time that an analyzer started a scan, relative to the
          // start of the MS run." [PSI:MS]
 
 
          spectrum_identification_result.retentionTime = cvParam.value.toDouble();
        }
    }
  if((spectrum_identification_result.scanNum == 0) &&
     (spectrum_identification_result.isSpectrumIndex == false))
    {
      m_qxmlStreamReader.raiseError(QObject::tr("scan number or spectrum index not found in "
                                                "SpectrumIdentificationResult id %1")
                                      .arg(spectrum_identification_result.id));
    }
 
 
  if(spectrum_identification_result.retentionTime == 0)
    {
      m_qxmlStreamReader.raiseError(
        QObject::tr("retention time not found in SpectrumIdentificationResult id %1")
          .arg(spectrum_identification_result.id));
    }
 
 
  for(auto spectrumIdentificationItem :
      spectrum_identification_result.spectrumIdentificationItemList)
    {
      // processSpectrumIdentificationItem(spectrum_identification_result,
      // spectrumIdentificationItem);
    }
}

Referenced by readAnalysisData().

◆ readStream()

void pappso::cbor::psm::MzIdentMlReader::readStream ( )

overrideprotectedvirtual

Implements pappso::XmlStreamReaderInterface.

Definition at line 194 of file mzidentmlreader.cpp.

{
  // mp_monitor->setStatus("reading X!Tandem result file");
  if(m_qxmlStreamReader.readNextStartElement())
    {
 
      qDebug() << m_qxmlStreamReader.name().toString();
      if(m_qxmlStreamReader.name().toString().toLower() == "mzidentml")
        {
          while(m_qxmlStreamReader.readNextStartElement())
            {
              // cvList
              if(m_qxmlStreamReader.name().toString() == "cvList")
                {
                  m_qxmlStreamReader.skipCurrentElement();
                }
              // AnalysisSoftwareList
              else if(m_qxmlStreamReader.name().toString() == "AnalysisSoftwareList")
                {
                  while(readAnalysisSoftware())
                    {
                    }
                } // Provider
              else if(m_qxmlStreamReader.name().toString() == "Provider")
                {
                  m_qxmlStreamReader.skipCurrentElement();
                }
              // AuditCollection
              else if(m_qxmlStreamReader.name().toString() == "AuditCollection")
                {
                  m_qxmlStreamReader.skipCurrentElement();
                }
              // SequenceCollection
              else if(m_qxmlStreamReader.name().toString() == "SequenceCollection")
                {
                  while(readSequenceCollectionItem())
                    {
                    }
                }
              // AnalysisCollection
              else if(m_qxmlStreamReader.name().toString() == "AnalysisCollection")
                {
                  while(readAnalysisCollectionItem())
                    {
                    }
                }
              // AnalysisProtocolCollection
              else if(m_qxmlStreamReader.name().toString() == "AnalysisProtocolCollection")
                {
                  readAnalysisProtocolCollection();
                }
 
              // DataCollection
              else if(m_qxmlStreamReader.name().toString() == "DataCollection")
                {
                  while(readDataCollectionItem())
                    {
                    }
                }
              else if(m_qxmlStreamReader.name().toString() == "BibliographicReference")
                {
                  m_qxmlStreamReader.skipCurrentElement();
                }
 
              else
                {
                  m_qxmlStreamReader.raiseError(
                    QObject::tr("element %1 not implemented").arg(m_qxmlStreamReader.name()));
                  m_qxmlStreamReader.skipCurrentElement();
                }
            }
        }
      else
        {
          m_qxmlStreamReader.raiseError(QObject::tr("Not an MzIdentML input file"));
          m_qxmlStreamReader.skipCurrentElement();
        }
    }
}

References pappso::XmlStreamReaderInterface::m_qxmlStreamReader, readAnalysisCollectionItem(), readAnalysisProtocolCollection(), readAnalysisSoftware(), readDataCollectionItem(), and readSequenceCollectionItem().

◆ readUserParam()

MzIdentMlReader::UserParam pappso::cbor::psm::MzIdentMlReader::readUserParam ( )

private

Definition at line 607 of file mzidentmlreader.cpp.

{
  UserParam user_param;
 
  user_param.name  = m_qxmlStreamReader.attributes().value("name").toString();
  user_param.value = m_qxmlStreamReader.attributes().value("value").toString();
  m_qxmlStreamReader.skipCurrentElement();
 
  return user_param;
}

References pappso::XmlStreamReaderInterface::m_qxmlStreamReader, pappso::cbor::psm::MzIdentMlReader::UserParam::name, and pappso::cbor::psm::MzIdentMlReader::UserParam::value.

Referenced by readSpectrumIdentificationItem(), and readSpectrumIdentificationResult().

◆ writeSpectrumIdentificationItem()

void pappso::cbor::psm::MzIdentMlReader::writeSpectrumIdentificationItem ( const SpectrumIdentificationItem & spectrum_identification_item )

private

Definition at line 1501 of file mzidentmlreader.cpp.

{
  mp_cborWriter->startMap(); // psm
  mp_cborWriter->append("proforma");
  mp_cborWriter->append(spectrum_identification_item.peptide.get()->toProForma());
 
  std::map<QString, std::vector<std::size_t>> map_protein_positions;
  for(auto &it_peptide_evidence : spectrum_identification_item.mzidPeptideEvidenceList)
    {
      auto it_insert =
        map_protein_positions.insert({it_peptide_evidence.protein.get()->getAccession(), {}});
      it_insert.first->second.push_back(it_peptide_evidence.start);
    }
  mp_cborWriter->append("protein_list");
  mp_cborWriter->startArray(map_protein_positions.size());
  for(auto &it_prot_pos : map_protein_positions)
    {
      mp_cborWriter->startMap();
      mp_cborWriter->append("accession");
      mp_cborWriter->append(it_prot_pos.first);
      mp_cborWriter->append("positions");
      mp_cborWriter->writeArray(it_prot_pos.second);
      mp_cborWriter->endMap();
    }
  mp_cborWriter->endArray();
 
 
  mp_cborWriter->append("eval");
  mp_cborWriter->startMap(); // start eval
  bool is_ok = false;
  is_ok      = writeTandemEval(spectrum_identification_item.cvParamList);
  mp_cborWriter->endMap(); // end eval
 
  if(!is_ok)
    {
      throw pappso::PappsoException("This identification engine is not taken into account");
    }
  mp_cborWriter->endMap(); // end psm
}

References pappso::cbor::psm::MzIdentMlReader::SpectrumIdentificationItem::cvParamList, mp_cborWriter, pappso::cbor::psm::MzIdentMlReader::SpectrumIdentificationItem::mzidPeptideEvidenceList, pappso::cbor::psm::MzIdentMlReader::SpectrumIdentificationItem::peptide, and writeTandemEval().

Referenced by writeSpectrumIdentificationResult().

◆ writeSpectrumIdentificationResult()

void pappso::cbor::psm::MzIdentMlReader::writeSpectrumIdentificationResult ( const SpectrumIdentificationResult & spectrum_identificatio_result )

private

Definition at line 1450 of file mzidentmlreader.cpp.

{
 
  mp_cborWriter->startMap(); // one scan
  mp_cborWriter->append("id");
  mp_cborWriter->startMap(); // id
  mp_cborWriter->append("index");
  mp_cborWriter->append((qint64)spectrum_identificatio_result.spectrumIndex);
  mp_cborWriter->append("native_id");
  mp_cborWriter->append(spectrum_identificatio_result.spectrumID);
  mp_cborWriter->endMap(); // end id
 
  mp_cborWriter->append("ms2");
  mp_cborWriter->startMap(); // ms2
  mp_cborWriter->append("rt");
  mp_cborWriter->append(spectrum_identificatio_result.retentionTime);
  mp_cborWriter->endMap(); // end ms2
 
 
  uint charge = spectrum_identificatio_result.spectrumIdentificationItemList.front().chargeState;
  double exp_mz =
    spectrum_identificatio_result.spectrumIdentificationItemList.front().experimentalMassToCharge;
 
  mp_cborWriter->append("precursor");
  mp_cborWriter->startMap(); // precursor
  mp_cborWriter->append("z");
  mp_cborWriter->append(charge);
  mp_cborWriter->append("mz");
  mp_cborWriter->append(exp_mz);
  mp_cborWriter->endMap(); // end precursor
 
 
  mp_cborWriter->append("psm_list");
  mp_cborWriter->startArray(spectrum_identificatio_result.spectrumIdentificationItemList.size());
  for(auto &spectrum_ident_item : spectrum_identificatio_result.spectrumIdentificationItemList)
    {
      if(spectrum_ident_item.chargeState != charge)
        { // error
        }
      if(spectrum_ident_item.experimentalMassToCharge != exp_mz)
        { // error
        }
      writeSpectrumIdentificationItem(spectrum_ident_item);
    }
  mp_cborWriter->endArray();
 
  mp_cborWriter->endMap(); // end one scan
}

References mp_cborWriter, pappso::cbor::psm::MzIdentMlReader::SpectrumIdentificationResult::retentionTime, pappso::cbor::psm::MzIdentMlReader::SpectrumIdentificationResult::spectrumID, pappso::cbor::psm::MzIdentMlReader::SpectrumIdentificationResult::spectrumIdentificationItemList, pappso::cbor::psm::MzIdentMlReader::SpectrumIdentificationResult::spectrumIndex, and writeSpectrumIdentificationItem().

Referenced by finalDebrief().

◆ writeTandemEval()

bool pappso::cbor::psm::MzIdentMlReader::writeTandemEval ( const std::vector< CvParam > & cv_param_list )

private

Definition at line 1544 of file mzidentmlreader.cpp.

{
  //        <cvParam accession="MS:1001330" cvRef="PSI-MS" name="X!Tandem:expect"
  //        value="0.0410883"/>
  //       <cvParam accession="MS:1001331" cvRef="PSI-MS" name="X!Tandem:hyperscore" value="25.9"/>
  auto it_find =
    std::find_if(cv_param_list.begin(), cv_param_list.end(), [](const CvParam &cv_param) {
      return cv_param.accession == "MS:1001330";
    });
  if(it_find != cv_param_list.end())
    {
      mp_cborWriter->append("xtandem");
      mp_cborWriter->startMap(); // start tandem
      mp_cborWriter->append("evalue");
      mp_cborWriter->append(it_find->value.toDouble());
      auto it_find =
        std::find_if(cv_param_list.begin(), cv_param_list.end(), [](const CvParam &cv_param) {
          return cv_param.accession == "MS:1001331";
        });
      mp_cborWriter->append("hyperscore");
      mp_cborWriter->append(it_find->value.toDouble());
 
      mp_cborWriter->endMap(); // end tandem
 
      return true;
    }
  return false;
}

References mp_cborWriter.

Referenced by writeSpectrumIdentificationItem().

Member Data Documentation

◆ m_analysisSoftwareVersion

QString pappso::cbor::psm::MzIdentMlReader::m_analysisSoftwareVersion

private

Definition at line 221 of file mzidentmlreader.h.

Referenced by readAnalysisSoftware().

◆ m_cborParameterMap

QCborMap pappso::cbor::psm::MzIdentMlReader::m_cborParameterMap

private

Definition at line 269 of file mzidentmlreader.h.

Referenced by finalDebrief(), and readAnalysisProtocolCollection().

◆ m_identificationEngine

IdentificationEngine pappso::cbor::psm::MzIdentMlReader::m_identificationEngine

private

Definition at line 220 of file mzidentmlreader.h.

Referenced by MzIdentMlReader().

◆ m_IdentificationEngineMap

std::map<QString, IdentificationEngine> pappso::cbor::psm::MzIdentMlReader::m_IdentificationEngineMap

private

store association between xml ID and an identification engine

Definition at line 228 of file mzidentmlreader.h.

Referenced by readAnalysisProtocolCollection(), and readAnalysisSoftware().

◆ m_MzidDBSequenceIdMap

std::map<QString, MzidDBSequence> pappso::cbor::psm::MzIdentMlReader::m_MzidDBSequenceIdMap

private

store association between xml ID and DBSequence

Definition at line 252 of file mzidentmlreader.h.

Referenced by finalDebrief(), readDBSequence(), and readPeptideEvidence().

◆ m_mzidentFile

QFileInfo pappso::cbor::psm::MzIdentMlReader::m_mzidentFile

private

Definition at line 268 of file mzidentmlreader.h.

Referenced by MzIdentMlReader(), and finalDebrief().

◆ m_MzidPeptideEvidenceIdMap

std::map<QString, MzidPeptideEvidence> pappso::cbor::psm::MzIdentMlReader::m_MzidPeptideEvidenceIdMap

private

store association between xml ID and peptide evidence

Definition at line 243 of file mzidentmlreader.h.

Referenced by readPeptideEvidence(), and readSpectrumIdentificationItem().

◆ m_mzidSearchDatabaseIdMap

std::map<QString, MzidSearchDatabase> pappso::cbor::psm::MzIdentMlReader::m_mzidSearchDatabaseIdMap

private

store association between xml ID and fasta files

Definition at line 233 of file mzidentmlreader.h.

Referenced by finalDebrief(), and readSearchDatabase().

◆ m_mzidSpectraDataIdMap

std::map<QString, MzidSpectraData> pappso::cbor::psm::MzIdentMlReader::m_mzidSpectraDataIdMap

private

store association between xml ID and SpectraData

Definition at line 248 of file mzidentmlreader.h.

Referenced by finalDebrief(), readSpectraData(), and readSpectrumIdentificationResult().

◆ m_PeptideIdMap

std::map<QString, PeptideSp> pappso::cbor::psm::MzIdentMlReader::m_PeptideIdMap

private

store association between xml ID and peptide sequence

Definition at line 238 of file mzidentmlreader.h.

Referenced by readPeptide(), readPeptideEvidence(), and readSpectrumIdentificationItem().

◆ m_proteinMap

PsmProteinMap pappso::cbor::psm::MzIdentMlReader::m_proteinMap

private

Definition at line 223 of file mzidentmlreader.h.

Referenced by finalDebrief().

◆ m_searchDatabase_ref2proteinList

std::map<QString, std::vector<ProteinSp> > pappso::cbor::psm::MzIdentMlReader::m_searchDatabase_ref2proteinList

private

associates database ref id to protein shared pointer because the search database id is not described before the protein (silly idea IMHO) we keep association of protein to database in this map until the real search database definition appears We then have to reprocess each protein to set the right fasta file pointer

Definition at line 260 of file mzidentmlreader.h.

◆ m_spectrumIdentificationResultBySpectraIdMap

std::map<QString, std::vector<SpectrumIdentificationResult> > pappso::cbor::psm::MzIdentMlReader::m_spectrumIdentificationResultBySpectraIdMap

private

store all identification results by spectra xml id

Definition at line 266 of file mzidentmlreader.h.

Referenced by finalDebrief(), and readSpectrumIdentificationResult().

◆ mp_cborWriter

pappso::cbor::CborStreamWriter* pappso::cbor::psm::MzIdentMlReader::mp_cborWriter

private

Definition at line 222 of file mzidentmlreader.h.

Referenced by MzIdentMlReader(), finalDebrief(), writeSpectrumIdentificationItem(), writeSpectrumIdentificationResult(), and writeTandemEval().

◆ mp_monitor

pappso::UiMonitorInterface* pappso::cbor::psm::MzIdentMlReader::mp_monitor

private

Definition at line 218 of file mzidentmlreader.h.

Referenced by MzIdentMlReader().

The documentation for this class was generated from the following files:

pappsomspp/core/processing/cbor/psm/mzidentml/mzidentmlreader.h
pappsomspp/core/processing/cbor/psm/mzidentml/mzidentmlreader.cpp

Classes

Public Member Functions

Protected Member Functions

Private Types

Private Member Functions

Private Attributes

Additional Inherited Members

Detailed Description

Member Enumeration Documentation

◆ IdentificationEngine

Constructor & Destructor Documentation

◆ MzIdentMlReader()

◆ ~MzIdentMlReader()

Member Function Documentation

◆ finalDebrief()

◆ readAnalysisCollectionItem()

◆ readAnalysisData()

◆ readAnalysisProtocolCollection()

◆ readAnalysisSoftware()

◆ readCvParam()

◆ readDataCollectionItem()

◆ readDBSequence()

◆ readInputs()

◆ readPeptide()

◆ readPeptideEvidence()

◆ readSearchDatabase()

◆ readSequenceCollectionItem()

◆ readSpectraData()

◆ readSpectrumIdentification()

◆ readSpectrumIdentificationItem()

◆ readSpectrumIdentificationResult()

◆ readStream()

◆ readUserParam()

◆ writeSpectrumIdentificationItem()

◆ writeSpectrumIdentificationResult()

◆ writeTandemEval()

Member Data Documentation

◆ m_analysisSoftwareVersion

◆ m_cborParameterMap

◆ m_identificationEngine

◆ m_IdentificationEngineMap

◆ m_MzidDBSequenceIdMap

◆ m_mzidentFile

◆ m_MzidPeptideEvidenceIdMap

◆ m_mzidSearchDatabaseIdMap

◆ m_mzidSpectraDataIdMap

◆ m_PeptideIdMap

◆ m_proteinMap

◆ m_searchDatabase_ref2proteinList

◆ m_spectrumIdentificationResultBySpectraIdMap

◆ mp_cborWriter

◆ mp_monitor