PeptideIdentification.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: $
// --------------------------------------------------------------------------
 
#pragma once
 
#include <OpenMS/METADATA/PeptideHit.h>
#include <OpenMS/METADATA/MetaInfoInterface.h>
#include <OpenMS/METADATA/ProteinHit.h>
#include <OpenMS/METADATA/IdentifierMSRunMapper.h>
#include <OpenMS/CONCEPT/HashUtils.h>
#include <OpenMS/METADATA/USI.h>
 
#include <functional>
#include <string>
#include <map>
 
namespace OpenMS
{
    class ConsensusMap;
    class PeptideIdentification;
 
    using SpectrumIdentification = PeptideIdentification; // better name that might become the default in future version
 
  class OPENMS_DLLAPI PeptideIdentification :
    public MetaInfoInterface
  {
public:
 
    typedef PeptideHit HitType;
 
 
 
    PeptideIdentification();
    virtual ~PeptideIdentification() noexcept;
    PeptideIdentification(const PeptideIdentification&) = default;
    PeptideIdentification(PeptideIdentification&&) noexcept = default;
 
    PeptideIdentification& operator=(const PeptideIdentification&) = default;
    PeptideIdentification& operator=(PeptideIdentification&&) = default; // TODO: add noexcept (gcc 4.8 bug)
    bool operator==(const PeptideIdentification& rhs) const;
    bool operator!=(const PeptideIdentification& rhs) const;
 
    double getRT() const;
    void setRT(double rt);
    bool hasRT() const;
 
    double getMZ() const;
    void setMZ(double mz);
    bool hasMZ() const;
 
    const std::vector<PeptideHit>& getHits() const;
    std::vector<PeptideHit>& getHits();
    void insertHit(const PeptideHit& hit);
    void insertHit(PeptideHit&& hit);
    void setHits(const std::vector<PeptideHit>& hits);
    void setHits(std::vector<PeptideHit>&& hits);
 
    double getSignificanceThreshold() const;
    void setSignificanceThreshold(double value);
 
    const std::string& getScoreType() const;
    void setScoreType(const std::string& type);
 
    bool isHigherScoreBetter() const;
    void setHigherScoreBetter(bool value);
 
    const std::string& getIdentifier() const;
    void setIdentifier(const std::string& id);
 
    std::string getBaseName() const;
    void setBaseName(const std::string& base_name);
 
    const std::string getExperimentLabel() const;
    void setExperimentLabel(const std::string& type);
 
    // TODO make a mandatory data member, add to idXML schema, think about storing the
    //  extracted spectrum "number" only!
    std::string getSpectrumReference() const;
    void setSpectrumReference(const std::string& ref);
 
    // Returns a higher or lower comparator based on @p higher_score_better_
    static std::function<bool(const PeptideHit&, const PeptideHit&)> getScoreComparator(bool higher_score_better);
 
    void sort();
 
    bool empty() const;
 
    static std::vector<PeptideHit> getReferencingHits(const std::vector<PeptideHit>&, const std::set<std::string>& accession);
 
    static std::multimap<std::string, std::pair<Size, Size>> buildUIDsFromAllPepIDs(const ConsensusMap &cmap);
 
    static std::string buildUIDFromPepID(const PeptideIdentification& pep_id,
                                    const IdentifierMSRunMapper& mapping);
 
    USI buildUSI(const std::string& ms_run_name,
                 const std::string& dataset_id = "local",
                 bool include_interpretation = false) const;
 
    USI buildUSI(const IdentifierMSRunMapper& mapping,
                 const std::string& dataset_id = "local",
                 bool include_interpretation = false) const;
 
protected:
    std::string id_; 
    std::vector<PeptideHit> hits_; 
    std::string score_type_; 
    bool higher_score_better_; 
    double mz_;
    double rt_;
  };
 
} //namespace OpenMS
 
// Hash function specialization for PeptideIdentification
namespace std
{
  template<>
  struct hash<OpenMS::PeptideIdentification>
  {
    std::size_t operator()(const OpenMS::PeptideIdentification& pi) const noexcept
    {
      // Start with MetaInfoInterface hash (includes significance_threshold, experiment_label, base_name)
      std::size_t seed = std::hash<OpenMS::MetaInfoInterface>{}(pi);
 
      // Hash identifier
      OpenMS::hash_combine(seed, OpenMS::fnv1a_hash_string(pi.getIdentifier()));
 
      // Hash all peptide hits
      for (const auto& hit : pi.getHits())
      {
        OpenMS::hash_combine(seed, std::hash<OpenMS::PeptideHit>{}(hit));
      }
 
      // Hash score type
      OpenMS::hash_combine(seed, OpenMS::fnv1a_hash_string(pi.getScoreType()));
 
      // Hash higher_score_better
      OpenMS::hash_combine(seed, OpenMS::hash_int(static_cast<int>(pi.isHigherScoreBetter())));
 
      // Hash mz (with NaN handling - NaN values hash to same value)
      if (pi.hasMZ())
      {
        OpenMS::hash_combine(seed, OpenMS::hash_float(pi.getMZ()));
      }
      else
      {
        // Hash a sentinel value for NaN (consistent with operator== which treats NaN==NaN as true)
        OpenMS::hash_combine(seed, OpenMS::hash_int(0xDEADBEEF));
      }
 
      // Hash rt (with NaN handling)
      if (pi.hasRT())
      {
        OpenMS::hash_combine(seed, OpenMS::hash_float(pi.getRT()));
      }
      else
      {
        // Hash a different sentinel value for NaN RT
        OpenMS::hash_combine(seed, OpenMS::hash_int(0xCAFEBABE));
      }
 
      return seed;
    }
  };
} // namespace std