ProteinResolver.h

Go to the documentation of this file.

// Copyright (c) 2002-2023, The OpenMS Team -- EKU Tuebingen, ETH Zurich, and FU Berlin
// SPDX-License-Identifier: BSD-3-Clause
//
// --------------------------------------------------------------------------
// $Maintainer: Timo Sachsenberg $
// $Authors: David Wojnar $
// --------------------------------------------------------------------------
 
#pragma once
 
#include <OpenMS/DATASTRUCTURES/DefaultParamHandler.h>
#include <OpenMS/KERNEL/ConsensusMap.h>
#include <OpenMS/KERNEL/FeatureMap.h>
#include <OpenMS/FORMAT/FASTAFile.h>
#include <OpenMS/METADATA/PeptideIdentification.h>
#include <OpenMS/METADATA/ProteinIdentification.h>
 
namespace OpenMS
{
  class OPENMS_DLLAPI ProteinResolver :
    public DefaultParamHandler
  {
 
public:
 
    //default constructor
    ProteinResolver();
 
    //copy constructor
    ProteinResolver(const ProteinResolver & rhs);
 
    //assignment operator
    ProteinResolver & operator=(const ProteinResolver & rhs);
 
    //destructor
    ~ProteinResolver() override;
 
 
    struct ProteinEntry;
    struct PeptideEntry;
    struct ISDGroup;
    struct MSDGroup;
    struct ResolverResult;
 
    struct ProteinEntry
    {
      std::list<PeptideEntry *> peptides;
      bool traversed;
      FASTAFile::FASTAEntry * fasta_entry;
      enum type  {primary, secondary, primary_indistinguishable, secondary_indistinguishable} protein_type;
      double weight;    //monoisotopic
      float coverage;    //in percent
      //if Protein is indistinguishable all his fellows are in the list indis
      std::list<ProteinEntry *> indis;
      Size index;
      Size  msd_group;     //index
      Size  isd_group;     //index
      Size number_of_experimental_peptides;
    };
 
    struct PeptideEntry
    {
      std::list<ProteinEntry *> proteins;
      bool traversed;
      String sequence;
      Size peptide_identification;
      Size peptide_hit;
      Size index;
      Size  msd_group;     //index
      Size isd_group;     //index
      bool experimental;
      float intensity;
      String origin;
    };
 
    struct MSDGroup
    {
      std::list<ProteinEntry *> proteins;
      std::list<PeptideEntry *> peptides;
      Size index;
      ISDGroup * isd_group;
      Size number_of_decoy;
      Size number_of_target;
      Size number_of_target_plus_decoy;
      float intensity;     
    };
 
    struct ISDGroup
    {
      std::list<ProteinEntry *> proteins;
      std::list<PeptideEntry *> peptides;
      Size index;
      std::list<Size> msd_groups;
    };
 
    struct ResolverResult
    {
      String identifier;
      std::vector<ISDGroup> * isds;
      std::vector<MSDGroup> * msds;
      std::vector<ProteinEntry> * protein_entries;
      std::vector<PeptideEntry> * peptide_entries;
      std::vector<Size> * reindexed_peptides;
      std::vector<Size> * reindexed_proteins;
      enum type  {PeptideIdent, Consensus} input_type;
      std::vector<PeptideIdentification> * peptide_identification;
      ConsensusMap * consensus_map;
    };
 
    void resolveConsensus(ConsensusMap & consensus);
 
    void resolveID(std::vector<PeptideIdentification> & peptide_identifications);
 
    // /**
      // @brief NOT IMPLEMENTED YET
 
      // @param protein_nodes
      // @param peptide_nodes
      // @param reindexed_proteins
      // @param reindexed_peptides
      // @param peptide_identifications
      // @param output
    // */
    // void writeProteinsAndPeptidesmzTab(std::vector<ProteinEntry>& protein_nodes, std::vector<PeptideEntry>& peptide_nodes, std::vector<Size>& reindexed_proteins, std::vector<Size>& reindexed_peptides, std::vector<PeptideIdentification>& peptide_identifications, String& output  );
    // /**
      // @brief Writing peptide table into text file
 
      // @param peptides
      // @param reindexed_peptides
      // @param identifications
      // @param output_file
    // */
    // void writePeptideTable(std::vector<PeptideEntry> & peptides, std::vector<Size> & reindexed_peptides, std::vector<PeptideIdentification> & identifications, String & output_file); // not implemented
    // /**
      // @brief Writing peptide table into text file
 
      // @param peptides
      // @param reindexed_peptides
      // @param consensus
      // @param output
    // */
    // void writePeptideTable(std::vector<PeptideEntry> & peptides, std::vector<Size> & reindexed_peptides, ConsensusMap & consensus, String & output_file); // not implemented
    // /**
      // @brief Writing protein table into text file
 
      // @param proteins
      // @param reindexed_proteins
      // @param output_file
    // */
    // void writeProteinTable(std::vector<ProteinEntry> & proteins, std::vector<Size> & reindexed_proteins, String & output_file); // not implemented
    // /**
      // @brief Writing protein groups into text file
 
      // @param isd_groups ISD groups
      // @param msd_groups MSD groups
      // @param output_file Path of output file
    // */
    // void writeProteinGroups(std::vector<ISDGroup> & isd_groups, std::vector<MSDGroup> & msd_groups, String & output_file); // not implemented
 
    void countTargetDecoy(std::vector<MSDGroup> & msd_groups, ConsensusMap & consensus);
 
    void countTargetDecoy(std::vector<MSDGroup> & msd_groups, std::vector<PeptideIdentification> & peptide_nodes);
 
    void clearResult();
 
    void setProteinData(std::vector<FASTAFile::FASTAEntry> & protein_data);
 
    const  std::vector<ResolverResult> & getResults();
 
    static const PeptideIdentification & getPeptideIdentification(const ConsensusMap & consensus, const PeptideEntry * peptide);
    static const PeptideHit & getPeptideHit(const ConsensusMap & consensus, const PeptideEntry * peptide);
    static const PeptideIdentification & getPeptideIdentification(const std::vector<PeptideIdentification> & peptide_nodes, const PeptideEntry * peptide);
    static const PeptideHit & getPeptideHit(const std::vector<PeptideIdentification> & peptide_nodes, const PeptideEntry * peptide);
 
private:
 
    std::vector<ResolverResult> resolver_result_;
    std::vector<FASTAFile::FASTAEntry> protein_data_;
 
    void computeIntensityOfMSD_(std::vector<MSDGroup> & msd_groups);
 
    void traverseProtein_(ProteinEntry * prot_node, MSDGroup & group);
    void traversePeptide_(PeptideEntry * pep_node, MSDGroup & group);
    Size findPeptideEntry_(String seq, std::vector<PeptideEntry> & nodes);
    Size binarySearchNodes_(String & seq, std::vector<PeptideEntry> & nodes, Size start, Size end);
    Size includeMSMSPeptides_(std::vector<PeptideIdentification> & peptide_identifications, std::vector<PeptideEntry> & peptide_nodes);
    Size includeMSMSPeptides_(ConsensusMap & consensus, std::vector<PeptideEntry> & peptide_nodes);
    void reindexingNodes_(std::vector<MSDGroup> & msd_groups, std::vector<Size> & reindexed_proteins, std::vector<Size> & reindexed_peptides);
    void primaryProteins_(std::vector<PeptideEntry> & peptide_nodes, std::vector<Size> & reindexed_peptides);
    void buildingMSDGroups_(std::vector<MSDGroup> & msd_groups, std::vector<ISDGroup> & isd_groups);
    void buildingISDGroups_(std::vector<ProteinEntry> & protein_nodes, std::vector<PeptideEntry> & peptide_nodes,
                            std::vector<ISDGroup> & isd_groups);
    // disabled/buggy
    //ProteinResolver::indistinguishableProteins(vector<MSDGroup>& msd_groups);
 
  }; // class
 
} // namespace