AASequence.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, Timo Sachsenberg $
// --------------------------------------------------------------------------
 
#pragma once
 
#include <OpenMS/CHEMISTRY/EmpiricalFormula.h>
#include <OpenMS/DATASTRUCTURES/StringUtils.h>
#include <OpenMS/CONCEPT/Types.h>
#include <OpenMS/CONCEPT/HashUtils.h>
#include <OpenMS/CHEMISTRY/Residue.h>
#include <OpenMS/CHEMISTRY/ResidueModification.h>
 
#include <vector>
#include <iosfwd>
#include <map>
#include <functional>
 
namespace OpenMS
{
 
 
 
 
  class OPENMS_DLLAPI AASequence final
  {
public:
 
    class Iterator;
 
    class OPENMS_DLLAPI ConstIterator final
    {
    public:
      // TODO Iterator constructor for ConstIterator
 
      typedef const Residue& const_reference;
      typedef Residue& reference;
      typedef const Residue* const_pointer;
      typedef std::vector<const Residue*>::difference_type difference_type;
      typedef Residue value_type;
      typedef const Residue* pointer;
      typedef std::random_access_iterator_tag iterator_category;
 
      ConstIterator() = default;
 
      ConstIterator(const std::vector<const Residue*>* vec_ptr, difference_type position)
        : vector_{vec_ptr},
          position_{position}
      {
      }
 
      ConstIterator(const ConstIterator& rhs) = default;
 
      ConstIterator(const AASequence::Iterator& rhs) :
        vector_(rhs.vector_),
        position_(rhs.position_)
      {
      }
 
      ~ConstIterator() = default;
 
 
      ConstIterator& operator=(const ConstIterator& rhs) = default;
 
      const_reference operator*() const
      {
        return *(*vector_)[position_];
      }
 
      const_pointer operator->() const
      {
        return (*vector_)[position_];
      }
 
      const ConstIterator operator+(difference_type diff) const
      {
        return ConstIterator(vector_, position_ + diff);
      }
 
      difference_type operator-(ConstIterator rhs) const
      {
        return position_ - rhs.position_;
      }
 
      const ConstIterator operator-(difference_type diff) const
      {
        return ConstIterator(vector_, position_ - diff);
      }
 
      bool operator==(const ConstIterator& rhs) const
      {
        return vector_ == rhs.vector_ && position_ == rhs.position_;
      }
 
      bool operator!=(const ConstIterator& rhs) const
      {
        return vector_ != rhs.vector_ || position_ != rhs.position_;
      }
 
      ConstIterator& operator++()
      {
        ++position_;
        return *this;
      }
 
      ConstIterator& operator--()
      {
        --position_;
        return *this;
      }
 
 
protected:
 
      // pointer to the AASequence vector
      const std::vector<const Residue*>* vector_ {};
 
      // position in the AASequence vector
      difference_type position_ {};
    };
 
 
    class OPENMS_DLLAPI Iterator final
    {
public:
 
      friend class AASequence::ConstIterator;
 
      typedef const Residue& const_reference;
      typedef Residue& reference;
      typedef const Residue* const_pointer;
      typedef const Residue* pointer;
      typedef std::vector<const Residue*>::difference_type difference_type;
 
      Iterator() = default;
 
      Iterator(std::vector<const Residue*>* vec_ptr, difference_type position)
        : vector_ {vec_ptr},
          position_{position}
      {
      }
 
      Iterator(const Iterator& rhs) = default;
 
      ~Iterator() = default;
 
 
      Iterator& operator=(const Iterator& rhs)
      {
        if (this != &rhs)
        {
          position_ = rhs.position_;
          vector_ = rhs.vector_;
        }
        return *this;
      }
 
      const_reference operator*() const
      {
        return *(*vector_)[position_];
      }
 
      const_pointer operator->() const
      {
        return (*vector_)[position_];
      }
 
      pointer operator->()
      {
        return (*vector_)[position_];
      }
 
      const Iterator operator+(difference_type diff) const
      {
        return Iterator(vector_, position_ + diff);
      }
 
      difference_type operator-(Iterator rhs) const
      {
        return position_ - rhs.position_;
      }
 
      const Iterator operator-(difference_type diff) const
      {
        return Iterator(vector_, position_ - diff);
      }
 
      bool operator==(const Iterator& rhs) const
      {
        return vector_ == rhs.vector_ && position_ == rhs.position_;
      }
 
      bool operator!=(const Iterator& rhs) const
      {
        return vector_ != rhs.vector_ || position_ != rhs.position_;
      }
 
      Iterator& operator++()
      {
        ++position_;
        return *this;
      }
 
      Iterator& operator--()
      {
        --position_;
        return *this;
      }
 
 
protected:
 
      // pointer to the AASequence vector
      std::vector<const Residue*>* vector_ {};
 
      // position in the AASequence vector
      difference_type position_ {};
    };
 
 
    AASequence() = default;
 
    AASequence(const AASequence&) = default;
 
    AASequence(AASequence&&) = default;
 
    ~AASequence() = default;
 
    AASequence& operator=(const AASequence&) = default;
 
    AASequence& operator=(AASequence&&) = default;
 
    bool empty() const;
 
 
    std::string toString() const;
 
    std::string toUnmodifiedString() const;
 
    std::string toUniModString() const;
 
    std::string toBracketString(bool integer_mass = true,
                           bool mass_delta = false,
                           const std::vector<std::string> & fixed_modifications = std::vector<std::string>()) const;
 
    void setModification(Size index, const std::string& modification);
 
    void setModification(Size index, const Residue* modification);
 
    void setModification(Size index, const ResidueModification* modification);
 
    void setModification(Size index, const ResidueModification& modification);
 
    void setModificationByDiffMonoMass(Size index, double diffMonoMass);
 
    void setNTerminalModification(const std::string& modification);
 
    void setNTerminalModification(const ResidueModification* modification);
 
    void setNTerminalModification(const ResidueModification& mod);
 
    void setNTerminalModificationByDiffMonoMass(double diffMonoMass, bool protein_term);
 
    const std::string& getNTerminalModificationName() const;
 
    const ResidueModification* getNTerminalModification() const;
 
    void setCTerminalModification(const std::string& modification);
 
    void setCTerminalModification(const ResidueModification* modification);
 
    void setCTerminalModification(const ResidueModification& mod);
 
    void setCTerminalModificationByDiffMonoMass(double diffMonoMass, bool protein_term);
 
    const std::string& getCTerminalModificationName() const;
 
    const ResidueModification* getCTerminalModification() const;
 
    const Residue& getResidue(Size index) const;
 
    EmpiricalFormula getFormula(Residue::ResidueType type = Residue::Full, Int charge = 0) const;
 
    double getAverageWeight(Residue::ResidueType type = Residue::Full, Int charge = 0) const;
 
    double getMonoWeight(Residue::ResidueType type = Residue::Full, Int charge = 0) const;
 
    double getMZ(Int charge, Residue::ResidueType type = Residue::Full) const;
 
    const Residue& operator[](Size index) const;
 
    AASequence operator+(const AASequence& peptide) const;
 
    AASequence& operator+=(const AASequence&);
 
    AASequence operator+(const Residue* residue) const;
 
    AASequence& operator+=(const Residue*);
 
    Size size() const;
 
    AASequence getPrefix(Size index) const;
 
    AASequence getSuffix(Size index) const;
 
    AASequence getSubsequence(Size index, UInt number) const;
 
    void getAAFrequencies(std::map<std::string, Size>& frequency_table) const;
 
 
    bool has(const Residue& residue) const;
 
    bool hasSubsequence(const AASequence& peptide) const;
 
    bool hasPrefix(const AASequence& peptide) const;
 
    bool hasSuffix(const AASequence& peptide) const;
 
    bool hasNTerminalModification() const;
 
    bool hasCTerminalModification() const;
 
    bool isModified() const;
 
    bool operator==(const AASequence& rhs) const;
 
    bool operator<(const AASequence& rhs) const;
 
    bool operator!=(const AASequence& rhs) const;
 
    inline Iterator begin() { return Iterator(&peptide_, 0); }
 
    inline ConstIterator begin() const { return ConstIterator(&peptide_, 0); }
 
    inline Iterator end() { return Iterator(&peptide_, (Int) peptide_.size()); }
 
    inline ConstIterator end() const { return ConstIterator(&peptide_, (Int) peptide_.size()); }
 
    friend OPENMS_DLLAPI std::ostream& operator<<(std::ostream& os, const AASequence& peptide);
 
    friend OPENMS_DLLAPI std::istream& operator>>(std::istream& is, const AASequence& peptide);
 
    static AASequence fromString(const std::string& s,
                                 bool permissive = true);
 
    static AASequence fromString(const char* s,
                                 bool permissive = true);
 
    explicit AASequence(const std::string& s);
 
    explicit AASequence(const char* s);
 
    explicit AASequence(const std::string& s, bool permissive);
 
    explicit AASequence(const char* s, bool permissive);
 
  protected:
    void validateSatelliteIon_(Residue::ResidueType type) const;
 
 
    std::vector<const Residue*> peptide_;
 
    const ResidueModification* n_term_mod_ = nullptr;
 
    const ResidueModification* c_term_mod_ = nullptr;
 
    static std::string::const_iterator parseModRoundBrackets_(const std::string::const_iterator str_it,
                                                        const std::string& str,
                                                        AASequence& aas,
                                                        const ResidueModification::TermSpecificity& specificity);
 
    static std::string::const_iterator parseModSquareBrackets_(const std::string::const_iterator str_it,
                                                               const std::string& str,
                                                               AASequence& aas,
                                                               const ResidueModification::TermSpecificity& specificity);
 
    static void parseString_(const std::string& peptide, AASequence& aas,
                             bool permissive = true);
  };
 
  OPENMS_DLLAPI std::ostream& operator<<(std::ostream& os, const AASequence& peptide);
 
  OPENMS_DLLAPI std::istream& operator>>(std::istream& os, const AASequence& peptide);
 
} // namespace OpenMS
 
// Hash function specialization for AASequence
// Placed in std namespace to allow use with std::unordered_map/set
namespace std
{
  template<>
  struct hash<OpenMS::AASequence>
  {
    std::size_t operator()(const OpenMS::AASequence& seq) const noexcept
    {
      std::size_t seed = 0;
 
      // Hash each residue
      for (const auto& residue : seq)
      {
        // Hash one-letter code (single character, fast)
        const std::string& olc = residue.getOneLetterCode();
        if (!olc.empty())
        {
          OpenMS::hash_combine(seed, OpenMS::hash_char(olc[0]));
        }
 
        // Hash modification if present
        const OpenMS::ResidueModification* mod = residue.getModification();
        if (mod != nullptr)
        {
          // Use full ID for portability (e.g., "Oxidation (M)")
          // std::string inherits from std::string, no copy needed
          OpenMS::hash_combine(seed, OpenMS::fnv1a_hash_string(mod->getFullId()));
        }
      }
 
      // Hash N-terminal modification if present
      const OpenMS::ResidueModification* n_mod = seq.getNTerminalModification();
      if (n_mod != nullptr)
      {
        // Use a different seed offset for N-term to distinguish from C-term
        std::size_t n_hash = OpenMS::fnv1a_hash_string(n_mod->getFullId());
        OpenMS::hash_combine(seed, n_hash ^ 0x4e5445524dULL); // "NTERM" in hex-like
      }
 
      // Hash C-terminal modification if present
      const OpenMS::ResidueModification* c_mod = seq.getCTerminalModification();
      if (c_mod != nullptr)
      {
        // Use a different seed offset for C-term
        std::size_t c_hash = OpenMS::fnv1a_hash_string(c_mod->getFullId());
        OpenMS::hash_combine(seed, c_hash ^ 0x435445524dULL); // "CTERM" in hex-like
      }
 
      return seed;
    }
  };
} // namespace std