Method mergePepXMLProtXML_

src/topp/IDMerger.cpp:73–148

protected:
  void mergePepXMLProtXML_(StringList filenames, vector<ProteinIdentification>&
                           proteins, vector<PeptideIdentification>& peptides)
  {
    IdXMLFile idxml;
    idxml.load(filenames[0], proteins, peptides);
    vector<ProteinIdentification> pepxml_proteins, protxml_proteins;
    vector<PeptideIdentification> pepxml_peptides, protxml_peptides;

    if (proteins[0].getProteinGroups().empty()) // first idXML contains data from the pepXML
    {
      proteins.swap(pepxml_proteins);
      peptides.swap(pepxml_peptides);
      idxml.load(filenames[1], protxml_proteins, protxml_peptides);
      if (protxml_proteins[0].getProteinGroups().empty())
      {
        throw Exception::InvalidParameter(__FILE__, __LINE__, OPENMS_PRETTY_FUNCTION, "None of the input files seems to be derived from a protXML file (information about protein groups is missing).");
      }
    }
    else // first idXML contains data from the protXML
    {
      proteins.swap(protxml_proteins);
      peptides.swap(protxml_peptides);
      idxml.load(filenames[1], pepxml_proteins, pepxml_peptides);
    }

    if ((protxml_peptides.size() > 1) || (protxml_proteins.size() > 1))
    {
      throw Exception::InvalidParameter(__FILE__, __LINE__, OPENMS_PRETTY_FUNCTION, "The idXML derived from a protXML file should contain only one 'ProteinIdentification' and one 'PeptideIdentification' instance.");
    }

    // peptide information comes from the pepXML (additional information in
    // the protXML - adapted peptide hit score, "is_unique", "is_contributing"
    // - is not transferred):
    peptides.swap(pepxml_peptides);

    // prepare scores and coverage values of protein hits from the protXML:
    map<String, pair<double, double> > hit_values;
    ProteinIdentification& protein = protxml_proteins[0];
    for (ProteinHit & hit : protein.getHits())
    {
      hit_values[hit.getAccession()] = make_pair(hit.getScore(), hit.getCoverage());
    }

    // merge protein information:
    proteins.swap(pepxml_proteins);
    for (ProteinIdentification & prot : proteins)
    {
      prot.getProteinGroups() = protein.getProteinGroups();
      prot.getIndistinguishableProteins() =
        protein.getIndistinguishableProteins();
      // TODO: since a protXML file can integrate data from several protein
      // identification runs, the protein groups/indistinguishable proteins
      // that we write to one identification run could contain references to
      // proteins that are not observed in this run, but in others; also, some
      // protein hits without enough evidence may not occur in the protXML
      // (thus also not in the protein groups) - clean this up?

      prot.setScoreType(protein.getScoreType());