msfileaccessor.cpp

Go to the documentation of this file.

// #include <proteowizard/pwiz/data/msdata/DefaultReaderList.hpp>
 
#include <QDebug>
#include <QFile>
#include <QFileInfo>
 
 
#include "msfileaccessor.h"
#include "pwizmsfilereader.h"
#include "pappsomspp/core/msfile/mzcbormsfilereader.h"
#include "pappsomspp/core/msfile/timsmsfilereader.h"
#include "pappsomspp/core/msfile/bafasciifilereader.h"
#include "pappsomspp/core/msfile/xymsfilereader.h"
 
#include "pappsomspp/core/exception/exceptionnotfound.h"
#include "pappsomspp/core/exception/exceptionnotpossible.h"
#include "pappsomspp/core/exception/exceptionnotrecognized.h"
#include "pappsomspp/core/msrun/msrunid.h"
#include "pappsomspp/core/msrun/private/mzcbormsrunreader.h"
#include "pappsomspp/core/msrun/private/timsframesmsrunreader.h"
 
#include "pappsomspp/core/msrun/private/pwizmsrunreader.h"
#include "pappsomspp/core/msrun/private/timsmsrunreader.h"
#include "pappsomspp/core/msrun/private/timsmsrunreaderms2.h"
#include "pappsomspp/core/msrun/private/timsmsrunreaderdia.h"
#include "pappsomspp/core/msrun/bafasciimsrunreader.h"
#include "pappsomspp/core/msrun/xymsrunreader.h"
 
#include "pappsomspp/core/utils.h"
 
 
namespace pappso
{
 
 
MsFileAccessor::MsFileAccessor(const QString &file_name, const QString &xml_prefix)
  : m_fileName(file_name), m_xmlPrefix(xml_prefix)
{
  QFile file(file_name);
  if(!file.exists())
    throw(ExceptionNotFound(
      QObject::tr("File %1 not found.").arg(QFileInfo(file_name).absoluteFilePath())));
 
 
  m_oboPsiModTermNativeIDFormat.setAccession("MS:1000824");
  m_oboPsiModTermNativeIDFormat.m_name = "no nativeID format";
  m_oboPsiModTermNativeIDFormat.m_definition =
    "No nativeID format indicates that the file tagged with this term does not "
    "contain spectra that can have a nativeID format.";
}
 
 
MsFileAccessor::MsFileAccessor(const MsFileAccessor &other)
  : m_fileName(other.m_fileName),
    m_xmlPrefix(other.m_xmlPrefix),
    m_fileFormat(other.m_fileFormat),
    m_fileReaderType(other.m_fileReaderType)
{
  m_oboPsiModTermNativeIDFormat = other.m_oboPsiModTermNativeIDFormat;
}
 
MsFileAccessor::~MsFileAccessor()
{
}
 
 
const QString &
MsFileAccessor::getFileName() const
{
  return m_fileName;
}
 
 
Enums::MsDataFormat
MsFileAccessor::getFileFormat() const
{
  return m_fileFormat;
}
 
const OboPsiModTerm
MsFileAccessor::getOboPsiModTermFileFormat() const
{
  OboPsiModTerm term;
 
  // is_a: MS:1000560 ! mass spectrometer file format
  switch(m_fileFormat)
    {
      case Enums::MsDataFormat::abSciexT2D:
        term.setAccession("MS:1001560");
        term.m_name = "SCIEX TOF/TOF T2D format";
        term.m_definition =
          "Applied Biosystems/MDS Analytical Technologies TOF/TOF instrument "
          "export format.";
        break;
      case Enums::MsDataFormat::abSciexWiff:
        term.setAccession("MS:1000562");
        term.m_name       = "ABI WIFF format";
        term.m_definition = "Applied Biosystems WIFF file format.";
        break;
      case Enums::MsDataFormat::agilentMassHunter:
        term.setAccession("MS:1001509");
        term.m_name = "Agilent MassHunter format";
        term.m_definition =
          "A data file format found in an Agilent MassHunter directory which "
          "contains raw data acquired by an Agilent mass spectrometer.";
        break;
      case Enums::MsDataFormat::brukerBaf:
        break;
      case Enums::MsDataFormat::brukerFid:
        term.setAccession("MS:1000825");
        term.m_name       = "Bruker FID format";
        term.m_definition = "Bruker FID file format.";
        break;
      case Enums::MsDataFormat::brukerTims:
        term.setAccession("MS:1002817");
        term.m_name       = "Bruker TDF format";
        term.m_definition = "Bruker TDF raw file format.";
        break;
      case Enums::MsDataFormat::brukerYep:
        term.setAccession("MS:1000567");
        term.m_name       = "Bruker/Agilent YEP format";
        term.m_definition = "Bruker/Agilent YEP file format.";
        break;
      case Enums::MsDataFormat::MGF:
        term.setAccession("MS:1001062");
        term.m_name       = "Mascot MGF format";
        term.m_definition = "Mascot MGF file format.";
        break;
      case Enums::MsDataFormat::msn:
        break;
      case Enums::MsDataFormat::mz5:
        term.setAccession("MS:1001881");
        term.m_name       = "mz5 format";
        term.m_definition = "mz5 file format, modelled after mzML.";
        break;
      case Enums::MsDataFormat::mzML:
        term.setAccession("MS:1000584");
        term.m_name       = "mzML format";
        term.m_definition = "Proteomics Standards Inititative mzML file format.";
        break;
      case Enums::MsDataFormat::mzcbor:
        // mzCBOR is a direct translation of mzML
        // waiting for a true OBO term, we choose the mzML definition
        term.setAccession("MS:1000584");
        term.m_name       = "mzML format";
        term.m_definition = "Proteomics Standards Inititative mzML file format.";
        break;
      case Enums::MsDataFormat::mzXML:
        term.setAccession("MS:1000566");
        term.m_name       = "ISB mzXML format";
        term.m_definition = "Institute of Systems Biology mzXML file format.";
        break;
      case Enums::MsDataFormat::SQLite3:
        break;
      case Enums::MsDataFormat::thermoRaw:
 
        term.setAccession("MS:1000563");
        term.m_name       = "Thermo RAW format";
        term.m_definition = "Thermo Scientific RAW file format.";
        break;
      case Enums::MsDataFormat::unknown:
        break;
      case Enums::MsDataFormat::watersRaw:
        term.setAccession("MS:1000526");
        term.m_name = "Waters raw format";
        term.m_definition =
          "Waters data file format found in a Waters RAW directory, generated "
          "from an MS acquisition.";
        break;
      case Enums::MsDataFormat::brukerBafAscii:
        term.setAccession("MS:1001369");
        term.m_name = "BafAscii text format";
        term.m_definition =
          "Simple text file format obtained by exporting Bruker Baf to ascii "
          "using Bruker software";
        break;
      case Enums::MsDataFormat::xy:
        term.setAccession("MS:1001369");
        term.m_name = "text format";
        term.m_definition =
          "Simple text file format of \"m/z<separator>intensity\" value pairs "
          "for a single mass spectrum, a PMF (or single MS2) search.";
        break;
      default:
        break;
    }
 
  return term;
}
 
 
const OboPsiModTerm &
MsFileAccessor::getOboPsiModTermNativeIDFormat() const
{
 
 
  return m_oboPsiModTermNativeIDFormat;
}
 
 
std::vector<MsRunIdCstSPtr>
MsFileAccessor::getMsRunIds()
{
  qDebug();
  // if (mzcbor_ms_file_reader.accept(m_fileName)) {
  // }
  std::vector<MsRunIdCstSPtr> ms_run_ids;
 
 
  // try the mzcbor file reader
  MzcborMsFileReader mzcbor_ms_file_reader(m_fileName);
  if(mzcbor_ms_file_reader.getFileFormat() == pappso::Enums::MsDataFormat::mzcbor)
    {
      qDebug() << "sure, this is mzcbor";
 
      ms_run_ids   = mzcbor_ms_file_reader.getMsRunIds(m_xmlPrefix);
      m_fileFormat = Enums::MsDataFormat::mzcbor;
      return ms_run_ids;
    }
 
  // Try the PwizMsFileReader
 
  PwizMsFileReader pwiz_ms_file_reader(m_fileName);
 
  ms_run_ids = pwiz_ms_file_reader.getMsRunIds(m_xmlPrefix);
  if(ms_run_ids.size())
    {
      qDebug() << "Might well be handled using the Pwiz code.";
 
      m_fileFormat     = pwiz_ms_file_reader.getFileFormat();
      m_fileReaderType = Enums::FileReaderType::pwiz;
 
      // But the user might have configured one preferred reader type.
 
      auto pref = m_preferredFileReaderTypeMap.find(m_fileFormat);
      if(pref != m_preferredFileReaderTypeMap.end())
        {
          m_fileReaderType = pref->second;
        }
 
      return ms_run_ids;
    }
 
  qDebug() << "The Pwiz reader did not work.";
 
  // Try the TimsData reader
 
  try
    {
      QString tims_dir = m_fileName;
      if(!QFileInfo(tims_dir).isDir())
        {
          tims_dir = QFileInfo(m_fileName).absolutePath();
        }
 
      TimsMsFileReader tims_file_reader(tims_dir);
 
      ms_run_ids = tims_file_reader.getMsRunIds(m_xmlPrefix);
 
      if(ms_run_ids.size())
        {
          qDebug() << "Might well be handled using the Bruker code";
 
          m_fileName       = tims_dir;
          m_fileFormat     = tims_file_reader.getFileFormat();
          m_fileReaderType = Enums::FileReaderType::tims;
 
          auto pref = m_preferredFileReaderTypeMap.find(m_fileFormat);
          if(pref != m_preferredFileReaderTypeMap.end())
            {
              m_fileReaderType = pref->second;
            }
 
          qDebug() << "Returning Bruker::tims ms run(s)."
                   << "with preferred reader type:"
                   << Utils::fileReaderTypeAsString(m_fileReaderType);
 
          return ms_run_ids;
        }
    }
  catch(const pappso::ExceptionNotRecognized &error)
    {
      qDebug() << "The Tims reader did not work.";
    }
 
 
  // Try the Baf->ascii export format from Bruker Compass
 
  try
    {
      ms_run_ids.clear();
      BafAsciiFileReader baf_ascii_ms_file_reader(m_fileName);
 
      ms_run_ids = baf_ascii_ms_file_reader.getMsRunIds(m_xmlPrefix);
 
      if(ms_run_ids.size())
        {
          qDebug() << "Might well be handled using the BafAscii code";
 
          m_fileReaderType = Enums::FileReaderType::bafascii;
 
          m_fileFormat = baf_ascii_ms_file_reader.getFileFormat();
 
          if(m_fileFormat == Enums::MsDataFormat::unknown)
            {
              ms_run_ids.clear();
            }
          else
            {
              return ms_run_ids;
            }
        }
    }
  catch(const pappso::PappsoException &error)
    {
      qDebug() << "This is not a BafAscii code file" << error.qwhat();
    }
 
 
  qDebug() << "The BafAscii reader did not work.";
 
  // At this point try the XyMsFileReader
 
  XyMsFileReader xy_ms_file_reader(m_fileName);
 
  ms_run_ids = xy_ms_file_reader.getMsRunIds(m_xmlPrefix);
 
  if(ms_run_ids.size())
    {
      qDebug() << "Might well be handled using the XY code";
 
      m_fileReaderType = Enums::FileReaderType::xy;
 
      m_fileFormat = xy_ms_file_reader.getFileFormat();
 
      return ms_run_ids;
    }
 
  qDebug() << "The XY reader did not work.";
 
 
  return ms_run_ids;
}
 
 
void
MsFileAccessor::setPreferredFileReaderType(Enums::MsDataFormat format,
                                           Enums::FileReaderType reader_type)
{
  // qDebug();
 
  auto ret = m_preferredFileReaderTypeMap.insert(
    std::pair<Enums::MsDataFormat, Enums::FileReaderType>(format, reader_type));
 
  if(!ret.second)
    {
      // replace
      ret.first->second = reader_type;
    }
}
 
 
Enums::FileReaderType
MsFileAccessor::getpreferredFileReaderType(Enums::MsDataFormat format)
{
  // qDebug();
 
  auto ret = m_preferredFileReaderTypeMap.find(format);
 
  if(ret != m_preferredFileReaderTypeMap.end())
    {
      return ret->second;
    }
 
  return m_fileReaderType;
}
 
 
Enums::FileReaderType
MsFileAccessor::getFileReaderType() const
{
  return m_fileReaderType;
}
 
 
void
MsFileAccessor::setSelectedMsRunId(MsRunIdCstSPtr ms_run_id_csp)
{
  mcsp_selectedMsRunId = ms_run_id_csp;
}
 
 
MsRunIdCstSPtr
MsFileAccessor::getSelectedMsRunId() const
{
  return mcsp_selectedMsRunId;
}
 
TimsMsRunReaderMs2SPtr
MsFileAccessor::buildTimsMsRunReaderMs2SPtr()
{
  // try TimsData reader
  QString tims_dir = m_fileName;
  if(!QFileInfo(tims_dir).isDir())
    {
      tims_dir = QFileInfo(m_fileName).absolutePath();
    }
  TimsMsFileReader tims_file_reader(tims_dir);
 
  std::vector<MsRunIdCstSPtr> ms_run_ids = tims_file_reader.getMsRunIds(m_xmlPrefix);
 
  if(ms_run_ids.size())
    {
      // qDebug() << "Might well be handled using the Bruker code";
      m_fileReaderType = Enums::FileReaderType::tims_ms2;
      m_fileFormat     = tims_file_reader.getFileFormat();
      m_fileName       = tims_dir;
 
      return std::make_shared<TimsMsRunReaderMs2>(ms_run_ids.front());
    }
  else
    {
      throw(ExceptionNotPossible(
        QObject::tr("Unable to read mz data directory %1 with TimsTOF reader.").arg(tims_dir)));
    }
}
 
 
MsRunReaderSPtr
MsFileAccessor::msRunReaderSPtr(MsRunIdCstSPtr ms_run_id)
{
  qDebug();
 
  // We want to return a MsRunReader that accounts for the configuration that
  // the user might have set.
 
  if(m_fileName != ms_run_id->getFileName())
    throw(
      ExceptionNotPossible(QObject::tr("The MsRunId instance must have the name file name as the "
                                       "MsFileAccessor. %1 != %2")
                             .arg(m_fileName)
                             .arg(ms_run_id->getFileName())));
 
  if(m_fileFormat == Enums::MsDataFormat::mzcbor)
    {
      qDebug() << "Returning a MzcborMsRunReader.";
 
      return std::make_shared<MzcborMsRunReader>(ms_run_id);
    }
  else if(getpreferredFileReaderType(m_fileFormat) == Enums::FileReaderType::pwiz)
    {
      qDebug() << "Returning a PwizMsRunReader.";
      auto pwiz_reader              = std::make_shared<PwizMsRunReader>(ms_run_id);
      m_oboPsiModTermNativeIDFormat = pwiz_reader->getOboPsiModTermNativeIDFormat();
      return pwiz_reader;
    }
  else if(getpreferredFileReaderType(m_fileFormat) == Enums::FileReaderType::xy)
    {
      // qDebug() << "Returning a XyMsRunReader.";
 
      return std::make_shared<XyMsRunReader>(ms_run_id);
    }
  else if(getpreferredFileReaderType(m_fileFormat) == Enums::FileReaderType::tims)
    {
      qDebug() << "Returning a TimsMsRunReader Enums::FileReaderType::tims";
      return std::make_shared<TimsMsRunReader>(ms_run_id);
    }
  else if(getpreferredFileReaderType(m_fileFormat) == Enums::FileReaderType::tims_frames)
    {
      qDebug() << "Returning a TimsFramesMsRunReader Enums::FileReaderType::tims_frames.";
 
      return std::make_shared<TimsFramesMsRunReader>(ms_run_id);
    }
  else if(getpreferredFileReaderType(m_fileFormat) == Enums::FileReaderType::tims_ms2)
    {
      qDebug() << "Returning a TimsMsRunReaderMs2 Enums::FileReaderType::tims_ms2";
 
      return std::make_shared<TimsMsRunReaderMs2>(ms_run_id);
    }
  else if(getpreferredFileReaderType(m_fileFormat) == Enums::FileReaderType::tims_dia)
    {
      qDebug() << "Returning a Enums::FileReaderType::tims_dia";
 
      // qInfo() << "std::make_shared<TimsMsRunReaderDia>(ms_run_id);";
      return std::make_shared<TimsMsRunReaderDia>(ms_run_id);
    }
  else if(getpreferredFileReaderType(m_fileFormat) == Enums::FileReaderType::bafascii)
    {
      // qDebug() << "Returning a BafAsciiMsRunReader.";
 
      return std::make_shared<BafAsciiMsRunReader>(ms_run_id);
    }
  if(m_fileFormat == Enums::MsDataFormat::unknown)
    {
      if(ms_run_id.get()->getMsDataFormat() == Enums::MsDataFormat::xy)
        {
          return std::make_shared<XyMsRunReader>(ms_run_id);
        }
      else
        {
          auto pwiz_reader              = std::make_shared<PwizMsRunReader>(ms_run_id);
          m_oboPsiModTermNativeIDFormat = pwiz_reader->getOboPsiModTermNativeIDFormat();
          return pwiz_reader;
        }
    }
  else
    {
      throw PappsoException(QObject::tr("No file format was found."));
    }
 
  return nullptr;
}
 
 
MsRunReaderSPtr
MsFileAccessor::msRunReaderSPtr(std::size_t ms_run_id_index)
{
  std::vector<MsRunIdCstSPtr> ms_run_ids = getMsRunIds();
  if(ms_run_id_index >= ms_run_ids.size())
    throw PappsoException(QObject::tr("MsRunId request out-of-bound error."));
 
  return msRunReaderSPtr(ms_run_ids.at(ms_run_id_index));
}
 
 
MsRunReaderSPtr
MsFileAccessor::msRunReaderSPtrForSelectedMsRunId()
{
  // qDebug();
 
  return msRunReaderSPtr(mcsp_selectedMsRunId);
}
 
 
MsRunReaderSPtr
MsFileAccessor::buildMsRunReaderSPtr(MsRunIdCstSPtr ms_run_id)
{
  return buildMsRunReaderSPtr(ms_run_id, Enums::FileReaderType::tims);
}
 
MsRunReaderSPtr
MsFileAccessor::buildMsRunReaderSPtr(MsRunIdCstSPtr ms_run_id,
                                     Enums::FileReaderType preferred_file_reader_type)
{
  QFile file(ms_run_id.get()->getFileName());
  if(!file.exists())
    throw(ExceptionNotFound(QObject::tr("unable to build a reader : file %1 not found.")
                              .arg(QFileInfo(ms_run_id.get()->getFileName()).absoluteFilePath())));
 
  Enums::MsDataFormat file_format = ms_run_id.get()->getMsDataFormat();
 
  if(file_format == Enums::MsDataFormat::xy)
    {
      // qDebug() << "Returning a XyMsRunReader.";
 
      return std::make_shared<XyMsRunReader>(ms_run_id);
    }
  else if(file_format == Enums::MsDataFormat::brukerBafAscii)
    {
      // qDebug() << "Returning a XyMsRunReader.";
 
      return std::make_shared<BafAsciiMsRunReader>(ms_run_id);
    }
  else if(file_format == Enums::MsDataFormat::unknown)
    {
      throw(PappsoException(QObject::tr("unable to build a reader for %1 : unknown file format")
                              .arg(QFileInfo(ms_run_id.get()->getFileName()).absoluteFilePath())));
    }
 
  else if(file_format == Enums::MsDataFormat::brukerTims)
    {
      if(preferred_file_reader_type == Enums::FileReaderType::tims)
        {
          return std::make_shared<TimsMsRunReader>(ms_run_id);
        }
      else if(preferred_file_reader_type == Enums::FileReaderType::tims_ms2)
        {
          return std::make_shared<TimsMsRunReaderMs2>(ms_run_id);
        }
      else if(preferred_file_reader_type == Enums::FileReaderType::tims_frames)
        {
          qDebug() << "returning std::make_shared<TimsFramesMsRunReader>(ms_run_id).";
          return std::make_shared<TimsFramesMsRunReader>(ms_run_id);
        }
      // qDebug() << "by default, build a TimsMsRunReader.";
      return std::make_shared<TimsMsRunReader>(ms_run_id);
    }
  else if(file_format == Enums::MsDataFormat::mzcbor)
    {
      return std::make_shared<MzcborMsRunReader>(ms_run_id);
    }
  else
    {
      // qDebug() << "Returning a PwizMsRunReader .";
      return std::make_shared<PwizMsRunReader>(ms_run_id);
    }
}
 
 
MsRunReaderSPtr
MsFileAccessor::getMsRunReaderSPtrByRunId(const QString &run_id, const QString &xml_id)
{
  std::vector<MsRunIdCstSPtr> run_list = getMsRunIds();
  MsRunReaderSPtr reader_sp;
  for(MsRunIdCstSPtr &original_run_id : run_list)
    {
      if(original_run_id.get()->getRunId() == run_id)
        {
          MsRunId new_run_id(*original_run_id.get());
          new_run_id.setXmlId(xml_id);
 
          return msRunReaderSPtr(std::make_shared<MsRunId>(new_run_id));
        }
    }
 
  if((run_id.isEmpty()) && (run_list.size() == 1))
    {
      MsRunId new_run_id(*run_list[0].get());
      new_run_id.setXmlId(xml_id);
 
      return msRunReaderSPtr(std::make_shared<MsRunId>(new_run_id));
    }
 
 
  if(reader_sp == nullptr)
    {
      throw(ExceptionNotFound(QObject::tr("run id %1 not found in file %2")
                                .arg(run_id)
                                .arg(QFileInfo(m_fileName).absoluteFilePath())));
    }
  return reader_sp;
}
 
 
} // namespace pappso