MzIdentMLHandler.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: Mathias Walzer $
// $Authors: Mathias Walzer, Andreas Bertsch $
// --------------------------------------------------------------------------
 
#pragma once
 
#include <OpenMS/KERNEL/StandardTypes.h>
 
#include <OpenMS/FORMAT/HANDLERS/XMLHandler.h>
#include <OpenMS/FORMAT/ControlledVocabulary.h>
#include <OpenMS/METADATA/ProteinHit.h>
#include <OpenMS/CHEMISTRY/AASequence.h>
#include <OpenMS/CHEMISTRY/DigestionEnzymeProtein.h>
#include <OpenMS/METADATA/ProteinIdentification.h>
#include <OpenMS/METADATA/PeptideIdentification.h>
 
#include <vector>
#include <map>
 
namespace OpenMS
{
  class ProgressLogger;
 
  namespace Internal
  {
 
  class OPENMS_DLLAPI IdentificationHit :
    public MetaInfoInterface
  {
  public:
 
    IdentificationHit() = default;
    
    IdentificationHit(const IdentificationHit&) = default;
    
    virtual ~IdentificationHit() = default;
    
    IdentificationHit(IdentificationHit&&) noexcept = default;
    
    IdentificationHit& operator=(const IdentificationHit&) = default;
    
    IdentificationHit& operator=(IdentificationHit&&) noexcept = default;
 
 
    bool operator==(const IdentificationHit& rhs) const noexcept;
    
    bool operator!=(const IdentificationHit& rhs) const noexcept;
 
 
    void setId(const std::string& id) noexcept;
    
    const std::string& getId() const noexcept;
    
    void setCharge(int charge) noexcept;
    
    int getCharge() const noexcept;
    
    void setCalculatedMassToCharge(double mz) noexcept;
    
    double getCalculatedMassToCharge() const noexcept;
    
    void setExperimentalMassToCharge(double mz) noexcept;
    
    double getExperimentalMassToCharge() const noexcept;
    
    void setName(const std::string& name) noexcept;
    
    const std::string& getName() const noexcept;
    
    void setPassThreshold(bool pass) noexcept;
    
    bool getPassThreshold() const noexcept;
    
    void setRank(int rank) noexcept;
    
    int getRank() const noexcept;
 
  private:
    std::string id_;                              
    int charge_ = 0;                             
    double calculated_mass_to_charge_ = 0.0;     
    double experimental_mass_to_charge_ = 0.0;   
    std::string name_;                           
    bool pass_threshold_ = true;                 
    int rank_ = 0;                               
  };
 
  class OPENMS_DLLAPI SpectrumIdentification :
    public MetaInfoInterface
  {
public:
 
 
    SpectrumIdentification() = default;
    virtual ~SpectrumIdentification();
    SpectrumIdentification(const SpectrumIdentification &) = default;
    SpectrumIdentification(SpectrumIdentification&&) = default;
    SpectrumIdentification & operator=(const SpectrumIdentification &) = default;
    SpectrumIdentification& operator=(SpectrumIdentification&&) & = default;
    bool operator==(const SpectrumIdentification & rhs) const;
    bool operator!=(const SpectrumIdentification & rhs) const;
 
    // @name Accessors
    void setHits(const std::vector<IdentificationHit> & hits);
 
    void addHit(const IdentificationHit & hit);
 
    const std::vector<IdentificationHit> & getHits() const;
 
protected:
 
    String id_; 
    std::vector<IdentificationHit> hits_; 
  };
 
    class OPENMS_DLLAPI Identification :
      public MetaInfoInterface
    {
  public:
 
 
 
      Identification() = default;
      Identification(const Identification & source) = default;
      Identification(Identification&&) = default;
      virtual ~Identification();
 
      Identification & operator=(const Identification & source) = default;
      Identification& operator=(Identification&&) & = default;
 
      bool operator==(const Identification & rhs) const;
      bool operator!=(const Identification & rhs) const;
 
 
      void setCreationDate(const DateTime & date);
 
      const DateTime & getCreationDate() const;
 
      void setSpectrumIdentifications(const std::vector<SpectrumIdentification> & ids);
 
      void addSpectrumIdentification(const SpectrumIdentification & id);
 
      const std::vector<SpectrumIdentification> & getSpectrumIdentifications() const;
  protected:
      String id_; 
      DateTime creation_date_; 
      std::vector<SpectrumIdentification> spectrum_identifications_;
    };
 
    class OPENMS_DLLAPI MzIdentMLHandler :
      public XMLHandler
    {
public:
      MzIdentMLHandler(const std::vector<ProteinIdentification>& pro_id, const std::vector<PeptideIdentification>& pep_id, const String& filename, const String& version, const ProgressLogger& logger);
 
      MzIdentMLHandler(std::vector<ProteinIdentification>& pro_id, std::vector<PeptideIdentification>& pep_id, const String& filename, const String& version, const ProgressLogger& logger);
 
      ~MzIdentMLHandler() override;
 
 
      // Docu in base class
      void endElement(const XMLCh* const /*uri*/, const XMLCh* const /*local_name*/, const XMLCh* const qname) override;
 
      // Docu in base class
      void startElement(const XMLCh* const /*uri*/, const XMLCh* const /*local_name*/, const XMLCh* const qname, const xercesc::Attributes& attributes) override;
 
      // Docu in base class
      void characters(const XMLCh* const chars, const XMLSize_t length) override;
 
      //Docu in base class
      void writeTo(std::ostream& os) override;
 
protected:
      const ProgressLogger& logger_;
 
      ControlledVocabulary cv_;
      ControlledVocabulary unimod_;
 
      //~ PeakMap* ms_exp_;
 
      String tag_;
 
      Identification* id_;
      std::vector<ProteinIdentification>* pro_id_;
      std::vector<PeptideIdentification>* pep_id_;
 
      const Identification* cid_;
      const std::vector<ProteinIdentification>* cpro_id_;
      const std::vector<PeptideIdentification>* cpep_id_;
 
      SpectrumIdentification current_spectrum_id_;
 
      IdentificationHit current_id_hit_;
 
      void handleCVParam_(const String& parent_parent_tag, const String& parent_tag, const String& accession, /* const String& name, */ /* const String& value, */ const xercesc::Attributes& attributes, const String& cv_ref /* ,  const String& unit_accession="" */);
 
      void handleUserParam_(const String& parent_parent_tag, const String& parent_tag, const String& name, const String& type, const String& value);
 
      void writeMetaInfos_(String& s, const MetaInfoInterface& meta, UInt indent) const;
 
      ControlledVocabulary::CVTerm getChildWithName_(const String& parent_accession, const String& name) const;
 
      //void writeSourceFile_(std::ostream& os, const String& id, const SourceFile& software);
 
      void writeEnzyme_(String& s, const DigestionEnzymeProtein& enzy, UInt miss, UInt indent) const;
 
      void writeModParam_(String& s, const std::vector<String>& mod_names, bool fixed, UInt indent) const;
 
      void writeFragmentAnnotations_(String& s, const std::vector<PeptideHit::PeakAnnotation>& annotations, UInt indent, bool is_ppxl) const;
 
      String trimOpenMSfileURI(const String& file) const;
 
      void writePeptideHit(const PeptideHit& hit,
                                std::vector<PeptideIdentification>::const_iterator& it,
                                std::map<String, String>& pep_ids,
                                const String& cv_ns, std::set<String>& sen_set,
                                std::map<String, String>& sen_ids,
                                std::map<String, std::vector<String> >& pep_evis,
                                std::map<String, double>& pp_identifier_2_thresh,
                                String& sidres);
 
      void writeXLMSPeptideHit(const PeptideHit& hit,
                                std::vector<PeptideIdentification>::const_iterator& it,
                                const String& ppxl_linkid, std::map<String, String>& pep_ids,
                                const String& cv_ns, std::set<String>& sen_set,
                                std::map<String, String>& sen_ids,
                                std::map<String, std::vector<String> >& pep_evis,
                                std::map<String, double>& pp_identifier_2_thresh,
                                double ppxl_crosslink_mass,
                                std::map<String, String>& ppxl_specref_2_element,
                                String& sid, bool alpha_peptide);
 
private:
      MzIdentMLHandler();
      MzIdentMLHandler(const MzIdentMLHandler& rhs);
      MzIdentMLHandler& operator=(const MzIdentMLHandler& rhs);
      std::map<String, AASequence> pep_sequences_;
      std::map<String, String> pp_identifier_2_sil_; 
      std::map<String, String> sil_2_sdb_; 
      std::map<String, String> sil_2_sdat_; 
      std::map<String, String> ph_2_sdat_; 
      std::map<String, String> sil_2_sip_; 
      AASequence actual_peptide_;
      Int current_mod_location_;
      ProteinHit actual_protein_;
 
    };
  } // namespace Internal
} // namespace OpenMS