ResidueModification.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: Timo Sachsenberg $
// $Authors: Andreas Bertsch $
// --------------------------------------------------------------------------
//
 
#pragma once
 
#include <OpenMS/DATASTRUCTURES/StringUtils.h>
#include <OpenMS/CHEMISTRY/EmpiricalFormula.h>
#include <OpenMS/CONCEPT/HashUtils.h>
 
#include <functional>
#include <set>
 
namespace OpenMS
{
  // forward declaration
  class Residue;
 
  class OPENMS_DLLAPI ResidueModification
  {
public:
 
    enum TermSpecificity
    {
      ANYWHERE = 0,        
      C_TERM = 1,          
      N_TERM = 2,          
      PROTEIN_C_TERM = 3,  
      PROTEIN_N_TERM = 4,  
      NUMBER_OF_TERM_SPECIFICITY
    };
 
    enum SourceClassification
    {
      ARTIFACT = 0,           
      HYPOTHETICAL,           
      NATURAL,                
      POSTTRANSLATIONAL,      
      MULTIPLE,               
      CHEMICAL_DERIVATIVE,    
      ISOTOPIC_LABEL,         
      PRETRANSLATIONAL,       
      OTHER_GLYCOSYLATION,    
      NLINKED_GLYCOSYLATION,  
      AA_SUBSTITUTION,        
      OTHER,                  
      NONSTANDARD_RESIDUE,    
      COTRANSLATIONAL,        
      OLINKED_GLYCOSYLATION,  
      UNKNOWN,                
      NUMBER_OF_SOURCE_CLASSIFICATIONS
    };
 
 
    ResidueModification();
 
    ResidueModification(const ResidueModification&) = default;
 
    ResidueModification(ResidueModification&&) = default;
 
    virtual ~ResidueModification();
 
 
    ResidueModification& operator=(const ResidueModification&) = default;
 
    ResidueModification& operator=(ResidueModification&&) & = default;
 
    void setId(const std::string& id);
 
    const std::string& getId() const;
 
    void setFullId(const std::string& full_id = "");
 
 
    const std::string& getFullId() const;
 
    void setUniModRecordId(const Int& id);
 
    const Int& getUniModRecordId() const;
 
    const std::string getUniModAccession() const;
 
    void setPSIMODAccession(const std::string& id);
 
    const std::string& getPSIMODAccession() const;
 
    void setFullName(const std::string& full_name);
 
    const std::string& getFullName() const;
 
    void setName(const std::string& name);
 
    const std::string& getName() const;
 
    void setTermSpecificity(TermSpecificity term_spec);
 
    void setTermSpecificity(const std::string& name);
 
    TermSpecificity getTermSpecificity() const;
 
    std::string getTermSpecificityName(TermSpecificity term_spec = NUMBER_OF_TERM_SPECIFICITY) const;
 
    void setOrigin(char origin);
 
    char getOrigin() const;
 
    void setSourceClassification(const std::string& classification);
 
    void setSourceClassification(SourceClassification classification);
 
    SourceClassification getSourceClassification() const;
 
    std::string getSourceClassificationName(SourceClassification classification = NUMBER_OF_SOURCE_CLASSIFICATIONS) const;
 
    void setAverageMass(double mass);
 
    double getAverageMass() const;
 
    void setMonoMass(double mass);
 
    double getMonoMass() const;
 
    void setDiffAverageMass(double mass);
 
    double getDiffAverageMass() const;
 
    void setDiffMonoMass(double mass);
 
    double getDiffMonoMass() const;
 
    void setFormula(const std::string& composition);
 
    const std::string& getFormula() const;
 
    void setDiffFormula(const EmpiricalFormula& diff_formula);
 
    const EmpiricalFormula& getDiffFormula() const;
 
    void setSynonyms(const std::set<std::string>& synonyms);
 
    void addSynonym(const std::string& synonym);
 
    const std::set<std::string>& getSynonyms() const;
 
    void setNeutralLossDiffFormulas(const std::vector<EmpiricalFormula>& diff_formulas);
 
    const std::vector<EmpiricalFormula>& getNeutralLossDiffFormulas() const;
 
    void setNeutralLossMonoMasses(std::vector<double> mono_masses);
 
    std::vector<double> getNeutralLossMonoMasses() const;
 
    void setNeutralLossAverageMasses(std::vector<double> average_masses);
 
    std::vector<double> getNeutralLossAverageMasses() const;
 
    bool hasNeutralLoss() const;
 
    bool isUserDefined() const;
 
    bool operator==(const ResidueModification& modification) const;
 
    bool operator!=(const ResidueModification& modification) const;
 
    bool operator<(const ResidueModification& modification) const;
 
 
 
    static const ResidueModification* createUnknownFromMassString(const std::string& mod,
                                                                  const double mass,
                                                                  const bool delta_mass,
                                                                  const TermSpecificity specificity,
                                                                  const Residue* residue = nullptr);
 
    static const ResidueModification* combineMods(const ResidueModification* base,
                                                  const std::set<const ResidueModification*>& addons,
                                                  bool allow_unknown_masses = false,
                                                  const Residue* residue = nullptr);
 
    std::string toString() const;
 
    static std::string getDiffMonoMassString(const double diff_mono_mass);
 
    static std::string getDiffMonoMassWithBracket(const double diff_mono_mass);
 
    static std::string getMonoMassWithBracket(const double mono_mass);
 
protected:
    std::string id_;
 
    std::string full_id_;
 
    std::string psi_mod_accession_;
 
    // The UniMod record id (an integer)
    Int unimod_record_id_;
 
    std::string full_name_;
 
    std::string name_;
 
    TermSpecificity term_spec_;
 
    char origin_;
 
    SourceClassification classification_;
 
    double average_mass_;
 
    double mono_mass_;
 
    double diff_average_mass_;
 
    double diff_mono_mass_;
 
    std::string formula_;
 
    EmpiricalFormula diff_formula_;
 
    std::set<std::string> synonyms_;
 
    std::vector<EmpiricalFormula> neutral_loss_diff_formulas_;
 
    std::vector<double> neutral_loss_mono_masses_;
 
    std::vector<double> neutral_loss_average_masses_;
  };
} // namespace OpenMS
 
// Hash function specialization for ResidueModification
// Placed in std namespace to allow use with std::unordered_map/set
namespace std
{
  template<>
  struct hash<OpenMS::ResidueModification>
  {
    std::size_t operator()(const OpenMS::ResidueModification& mod) const noexcept
    {
      // Hash string fields
      std::size_t seed = OpenMS::fnv1a_hash_string(mod.getId());
      OpenMS::hash_combine(seed, OpenMS::fnv1a_hash_string(mod.getFullId()));
      OpenMS::hash_combine(seed, OpenMS::fnv1a_hash_string(mod.getPSIMODAccession()));
      OpenMS::hash_combine(seed, OpenMS::hash_int(mod.getUniModRecordId()));
      OpenMS::hash_combine(seed, OpenMS::fnv1a_hash_string(mod.getFullName()));
      OpenMS::hash_combine(seed, OpenMS::fnv1a_hash_string(mod.getName()));
 
      // Hash enum fields (as integers)
      OpenMS::hash_combine(seed, OpenMS::hash_int(static_cast<int>(mod.getTermSpecificity())));
      OpenMS::hash_combine(seed, OpenMS::hash_char(mod.getOrigin()));
      OpenMS::hash_combine(seed, OpenMS::hash_int(static_cast<int>(mod.getSourceClassification())));
 
      // Hash mass fields
      OpenMS::hash_combine(seed, OpenMS::hash_float(mod.getAverageMass()));
      OpenMS::hash_combine(seed, OpenMS::hash_float(mod.getMonoMass()));
      OpenMS::hash_combine(seed, OpenMS::hash_float(mod.getDiffAverageMass()));
      OpenMS::hash_combine(seed, OpenMS::hash_float(mod.getDiffMonoMass()));
 
      // Hash formula fields
      OpenMS::hash_combine(seed, OpenMS::fnv1a_hash_string(mod.getFormula()));
      OpenMS::hash_combine(seed, std::hash<OpenMS::EmpiricalFormula>{}(mod.getDiffFormula()));
 
      // Hash synonyms (set<std::string>)
      const auto& synonyms = mod.getSynonyms();
      OpenMS::hash_combine(seed, OpenMS::hash_int(synonyms.size()));
      for (const auto& syn : synonyms)
      {
        OpenMS::hash_combine(seed, OpenMS::fnv1a_hash_string(syn));
      }
 
      // Hash neutral loss diff formulas (vector<EmpiricalFormula>)
      const auto& nl_formulas = mod.getNeutralLossDiffFormulas();
      OpenMS::hash_combine(seed, OpenMS::hash_int(nl_formulas.size()));
      for (const auto& nlf : nl_formulas)
      {
        OpenMS::hash_combine(seed, std::hash<OpenMS::EmpiricalFormula>{}(nlf));
      }
 
      // Hash neutral loss mono masses (vector<double>)
      const auto& nl_mono = mod.getNeutralLossMonoMasses();
      OpenMS::hash_combine(seed, OpenMS::hash_int(nl_mono.size()));
      for (double mass : nl_mono)
      {
        OpenMS::hash_combine(seed, OpenMS::hash_float(mass));
      }
 
      // Hash neutral loss average masses (vector<double>)
      const auto& nl_avg = mod.getNeutralLossAverageMasses();
      OpenMS::hash_combine(seed, OpenMS::hash_int(nl_avg.size()));
      for (double mass : nl_avg)
      {
        OpenMS::hash_combine(seed, OpenMS::hash_float(mass));
      }
 
      return seed;
    }
  };
} // namespace std