PepXMLFile.h

Go to the documentation of this file.

// Copyright (c) 2002-present, OpenMS Inc. -- EKU Tuebingen, ETH Zurich, and FU Berlin
// SPDX-License-Identifier: BSD-3-Clause
//
// --------------------------------------------------------------------------
// $Maintainer: Chris Bielow, Hendrik Weisser $
// $Authors: Chris Bielow, Hendrik Weisser $
// --------------------------------------------------------------------------
 
#pragma once
 
#include <OpenMS/CHEMISTRY/AASequence.h>
#include <OpenMS/CHEMISTRY/Element.h>
#include <OpenMS/CHEMISTRY/ResidueDB.h>
#include <OpenMS/FORMAT/HANDLERS/XMLHandler.h>
#include <OpenMS/FORMAT/XMLFile.h>
#include <OpenMS/METADATA/PeptideIdentification.h>
#include <OpenMS/METADATA/PeptideIdentificationList.h>
#include <OpenMS/METADATA/ProteinIdentification.h>
#include <OpenMS/METADATA/SpectrumMetaDataLookup.h>
 
#include <vector>
#include <map>
#include <set>
 
 
namespace OpenMS
{
  class OPENMS_DLLAPI PepXMLFile :
    protected Internal::XMLHandler,
    public Internal::XMLFile
  {
public:
 
    PepXMLFile();
 
    ~PepXMLFile() override;
 
    void load(const std::string& filename,
              std::vector<ProteinIdentification>& proteins,
              PeptideIdentificationList& peptides,
              const std::string& experiment_name,
              const SpectrumMetaDataLookup& lookup);
 
    void load(const std::string& filename,
              std::vector<ProteinIdentification>& proteins,
              PeptideIdentificationList& peptides,
              const std::string& experiment_name = "");
 
    void store(const std::string& filename, std::vector<ProteinIdentification>& protein_ids,
               PeptideIdentificationList& peptide_ids, const std::string& mz_file = "",
               const std::string& mz_name = "", bool peptideprophet_analyzed = false, double rt_tolerance = 0.01);
 
    void keepNativeSpectrumName(bool keep)
    {
      keep_native_name_ = keep;
    }
 
    void setPreferredFixedModifications(const std::vector<const ResidueModification*>& mods);
 
    void setPreferredVariableModifications(const std::vector<const ResidueModification*>& mods);
 
    void setParseUnknownScores(bool parse_unknown_scores);
 
protected:
 
    void endElement(const XMLCh* const /*uri*/, const XMLCh* const /*local_name*/, const XMLCh* const qname) override;
 
    void startElement(const XMLCh* const /*uri*/, const XMLCh* const /*local_name*/, const XMLCh* const qname, const xercesc::Attributes& attributes) override;
 
private:
 
    void makeScanMap_();
 
    void readRTMZCharge_(const xercesc::Attributes& attributes);
 
    struct AminoAcidModification
    {
      private:
 
      std::string aminoacid_;
      double massdiff_;
      double mass_;
      bool is_variable_;
      std::string description_;
      std::string terminus_;
      bool is_protein_terminus_; // "true" if protein terminus, "false" if peptide terminus
      ResidueModification::TermSpecificity term_spec_;
      std::vector<std::string> errors_;
      const ResidueModification* registered_mod_;
 
      const ResidueModification* lookupModInPreferredMods_(const std::vector<const ResidueModification*>& preferred_fixed_mods,
                                                           const std::string& aminoacid,
                                                           double massdiff,
                                                           const std::string& description,
                                                           const ResidueModification::TermSpecificity term_spec,
                                                           double tolerance);
 
      public:
      AminoAcidModification() = delete;
 
      AminoAcidModification(
          const std::string& aminoacid, const std::string& massdiff, const std::string& mass,
          std::string variable, const std::string& description, std::string terminus, const std::string& protein_terminus,
          const std::vector<const ResidueModification*>& preferred_fixed_mods,
          const std::vector<const ResidueModification*>& preferred_var_mods,
          double tolerance);
 
      AminoAcidModification(const AminoAcidModification& rhs) = default;
 
      virtual ~AminoAcidModification() = default;
 
      AminoAcidModification& operator=(const AminoAcidModification& rhs) = default;
 
      std::string toUnimodLikeString() const;
 
      const std::string& getDescription() const;
 
      bool isVariable() const;
 
      const ResidueModification* getRegisteredMod() const;
 
      double getMassDiff() const;
 
      double getMass() const;
 
      const std::string& getTerminus() const;
 
      const std::string& getAminoAcid() const;
 
      const std::vector<std::string>& getErrors() const;
    };
 
    std::vector<ProteinIdentification>* proteins_;
 
    PeptideIdentificationList* peptides_;
 
    const SpectrumMetaDataLookup* lookup_;
 
    std::string exp_name_;
 
    std::string search_engine_;
 
    std::string native_spectrum_name_;
    std::string experiment_label_;
    std::string swath_assay_;
    std::string status_;
 
    bool use_precursor_data_{};
 
    std::map<Size, Size> scan_map_;
 
    Element hydrogen_;
 
    bool analysis_summary_;
 
    bool keep_native_name_;
 
    bool search_score_summary_;
 
    bool search_summary_{};
 
    bool wrong_experiment_{};
 
    bool seen_experiment_{};
 
    bool checked_base_name_{};
 
    bool has_decoys_{};
 
    bool parse_unknown_scores_{};
 
    std::string decoy_prefix_;
 
    std::string current_base_name_;
 
    std::vector<std::vector<ProteinIdentification>::iterator> current_proteins_;
 
    ProteinIdentification::SearchParameters params_;
 
    std::string enzyme_;
    std::string enzyme_cuttingsite_;
 
    PeptideIdentification current_peptide_;
 
    PeptideHit::PepXMLAnalysisResult current_analysis_result_;
 
    PeptideHit peptide_hit_;
 
    std::string current_sequence_;
 
    double rt_{}, mz_{};
 
    Size scannr_{};
 
    Int charge_{};
 
    UInt search_id_{};
 
    std::string prot_id_;
 
    DateTime date_;
 
    double hydrogen_mass_{};
 
    std::vector<std::pair<const ResidueModification*, Size> > current_modifications_;
 
    std::vector<AminoAcidModification> fixed_modifications_;
 
    std::vector<AminoAcidModification> variable_modifications_;
 
    std::vector<const ResidueModification*> preferred_fixed_modifications_;
 
    std::vector<const ResidueModification*> preferred_variable_modifications_;
 
 
    static const double mod_tol_;
    static const double xtandem_artificial_mod_tol_;
 
    bool lookupAddFromHeader_(double modification_mass,
                              Size modification_position,
                              std::vector<AminoAcidModification> const& header_mods);
 
    //static std::vector<int> getIsotopeErrorsFromIntSetting_(int intSetting);
  };
} // namespace OpenMS