Method load

src/openms/source/FORMAT/PercolatorOutfile.cpp:170–334

  void PercolatorOutfile::load(const String& filename,
                               ProteinIdentification& proteins,
                               vector<PeptideIdentification>& peptides,
                               SpectrumMetaDataLookup& lookup,
                               enum ScoreType output_score)
  {
    SpectrumMetaDataLookup::MetaDataFlags lookup_flags =
      (SpectrumMetaDataLookup::MDF_RT |
       SpectrumMetaDataLookup::MDF_PRECURSORMZ |
       SpectrumMetaDataLookup::MDF_PRECURSORCHARGE);

    if (lookup.reference_formats.empty())
    {
      // MS-GF+ Percolator (mzid?) format:
      lookup.addReferenceFormat(R"(_SII_(?<INDEX1>\d+)_\d+_\d+_(?<CHARGE>\d+)_\d+)");
      // Mascot Percolator format (RT may be missing, e.g. for searches via
      // ProteomeDiscoverer):
      lookup.addReferenceFormat(R"(spectrum:[^;]+[(scans:)(scan=)(spectrum=)](?<INDEX0>\d+)[^;]+;rt:(?<RT>\d*(\.\d+)?);mz:(?<MZ>\d+(\.\d+)?);charge:(?<CHARGE>-?\d+))");
      // X! Tandem Percolator format:
      lookup.addReferenceFormat(R"(_(?<INDEX0>\d+)_(?<CHARGE>\d+)_\d+$)");
    }

    vector<String> items;
    CsvFile source(filename, '\t');
    source.getRow(0, items);
    String header = ListUtils::concatenate<String>(items, '\t');
    if (header != 
        "PSMId\tscore\tq-value\tposterior_error_prob\tpeptide\tproteinIds")
    {
      throw Exception::ParseError(__FILE__, __LINE__, OPENMS_PRETTY_FUNCTION,
                                  header, "Not a valid header for Percolator (PSM level) output");
    }

    set<String> accessions;
    Size no_charge = 0, no_rt = 0, no_mz = 0; // counters for missing data
    peptides.clear();

    for (Size row = 1; row < source.rowCount(); ++row)
    {
      source.getRow(row, items);

      SpectrumMetaDataLookup::SpectrumMetaData meta_data;
      try
      {
        lookup.getSpectrumMetaData(items[0], meta_data, lookup_flags);
      }
      catch (...)
      {
        String msg = "Error: Could not extract data for spectrum reference '" +
          items[0] + "' from row " + String(row);
        OPENMS_LOG_ERROR << msg << endl;
      }

      PeptideHit hit;
      if (meta_data.precursor_charge != 0)
      {
        hit.setCharge(meta_data.precursor_charge);
      }