pappso::ProteinIntegerCode Class Reference

#include <proteinintegercode.h>

Public Member Functions
	ProteinIntegerCode (ProteinSp protein, const AaStringCodec &codec, std::size_t aa_str_max_size=5)
	ProteinIntegerCode (const ProteinIntegerCode &other)
virtual	~ProteinIntegerCode ()
const std::vector< std::uint8_t > &	getSeqAaCode () const
const std::vector< std::uint32_t > &	getPeptideCodedFragment (std::size_t size) const
std::vector< std::pair< std::size_t, std::uint32_t > >	match (const std::vector< uint32_t > &code_list) const
	list of positions and matched codes along protein sequence
std::vector< double >	convolution (const std::vector< uint32_t > &code_list_from_spectrum) const
	process convolution of spectrum code list along protein sequence
pappso::ProteinSp	getProteinSp () const

Private Member Functions
std::vector< std::uint32_t >	computePeptideCodeFragments (const AaStringCodec &codec, std::size_t fragment_size) const
double	convolutionKernel (const std::vector< uint32_t > &spectrum_code_list, std::vector< std::uint8_t >::const_iterator it_aa, std::vector< std::uint32_t >::const_iterator it_couple, std::vector< std::uint32_t >::const_iterator it_trio, std::vector< std::uint32_t >::const_iterator it_quatro, std::vector< std::uint32_t >::const_iterator it_cinqo) const

Private Attributes
ProteinSp	msp_protein
std::vector< std::uint8_t >	m_seqAaCode
std::vector< std::vector< std::uint32_t > >	m_peptideCodedFragments

Detailed Description

Definition at line 43 of file proteinintegercode.h.

Constructor & Destructor Documentation

ProteinIntegerCode() [1/2]

ProteinIntegerCode::ProteinIntegerCode	(	ProteinSp	protein,
		const AaStringCodec &	codec,
		std::size_t	aa_str_max_size = 5 )

Default constructor

Definition at line 35 of file proteinintegercode.cpp.

{
  msp_protein = protein;
 
  if(aa_str_max_size > 7)
    {
      throw ExceptionOutOfRange(
        QObject::tr("aa_str_max_size exceeds max size"));
    }
 
  QString seq_str = protein.get()->getSequence();
  m_seqAaCode.clear();
 
  for(const QChar &aa_str : seq_str)
    {
      m_seqAaCode.push_back(codec.getAaCode().getAaCode(aa_str.toLatin1()));
    }
 
  for(std::size_t i = 2; i <= aa_str_max_size; i++)
    {
      m_peptideCodedFragments.push_back(computePeptideCodeFragments(codec, i));
    }
}

References computePeptideCodeFragments(), pappso::AaCode::getAaCode(), pappso::AaStringCodec::getAaCode(), pappso::Protein::getSequence(), m_peptideCodedFragments, m_seqAaCode, and msp_protein.

Referenced by ProteinIntegerCode().

ProteinIntegerCode() [2/2]

ProteinIntegerCode::ProteinIntegerCode

(

const ProteinIntegerCode &

other

)

Copy constructor

Parameters

other

TODO

Definition at line 61 of file proteinintegercode.cpp.

{
 
  msp_protein = other.msp_protein;
 
  m_seqAaCode             = other.m_seqAaCode;
  m_peptideCodedFragments = other.m_peptideCodedFragments;
}

References ProteinIntegerCode(), m_peptideCodedFragments, m_seqAaCode, and msp_protein.

~ProteinIntegerCode()

ProteinIntegerCode::~ProteinIntegerCode ( )

virtual

Destructor

Definition at line 70 of file proteinintegercode.cpp.

{
}

Member Function Documentation

computePeptideCodeFragments()

std::vector< std::uint32_t > pappso::ProteinIntegerCode::computePeptideCodeFragments ( const AaStringCodec & codec, std::size_t fragment_size ) const

private

Definition at line 87 of file proteinintegercode.cpp.

{
  std::vector<std::uint32_t> fragments;
 
  int max = (m_seqAaCode.size() - fragment_size);
  if(max < 0)
    return fragments;
 
  auto it = m_seqAaCode.begin();
  for(int i = 0; i <= max; i++)
    {
      fragments.push_back(codec.codeLlc(it, fragment_size));
      it++;
    }
 
  return fragments;
}

References pappso::AaStringCodec::codeLlc(), and m_seqAaCode.

Referenced by ProteinIntegerCode().

convolution()

std::vector< double > pappso::ProteinIntegerCode::convolution

(

const std::vector< uint32_t > &

code_list_from_spectrum

)

const

process convolution of spectrum code list along protein sequence

Parameters

code_list_from_spectrum

unique sorted sequence codes given by the spectrum decoder

Definition at line 165 of file proteinintegercode.cpp.

{
  std::vector<double> convolution_score;
 
 
  std::vector<std::uint8_t>::const_iterator it_aa = m_seqAaCode.begin();
  auto it_couple = m_peptideCodedFragments[0].begin();
  auto it_trio   = m_peptideCodedFragments[1].begin();
  auto it_quatro = m_peptideCodedFragments[2].begin();
  auto it_cinqo  = m_peptideCodedFragments[3].begin();
  for(std::uint8_t aa_code [[maybe_unused]] : m_seqAaCode)
    {
      convolution_score.push_back(convolutionKernel(code_list_from_spectrum,
                                                    it_aa,
                                                    it_couple,
                                                    it_trio,
                                                    it_quatro,
                                                    it_cinqo));
      it_aa++;
      it_couple++;
      it_trio++;
      it_quatro++;
      it_cinqo++;
    }
 
  return convolution_score;
}

References convolutionKernel(), m_peptideCodedFragments, and m_seqAaCode.

convolutionKernel()

double pappso::ProteinIntegerCode::convolutionKernel ( const std::vector< uint32_t > & spectrum_code_list, std::vector< std::uint8_t >::const_iterator it_aa, std::vector< std::uint32_t >::const_iterator it_couple, std::vector< std::uint32_t >::const_iterator it_trio, std::vector< std::uint32_t >::const_iterator it_quatro, std::vector< std::uint32_t >::const_iterator it_cinqo ) const

private

Definition at line 195 of file proteinintegercode.cpp.

{
  double score = 0;
 
  // single :
  auto it_end         = it_aa + 5;
  auto it             = it_aa;
  double single_score = 0;
  while(it != it_end)
    {
      if(std::binary_search(spectrum_code_list.begin(),
                            spectrum_code_list.end(),
                            (std::uint32_t)(*it)))
        {
          single_score += 1;
        }
      else
        {
          single_score += 0.1;
        }
      it++;
    }
 
  // duo
  auto itduo_end   = it_couple + 4;
  auto itduo       = it_couple;
  double duo_score = 0;
  while(itduo != itduo_end)
    {
      if(std::binary_search(
           spectrum_code_list.begin(), spectrum_code_list.end(), *itduo))
        {
          duo_score += 3;
        }
      else
        {
          duo_score += 0.2;
        }
      itduo++;
    }
 
 
  // trio
  auto it3_end      = it_trio + 3;
  auto it3          = it_trio;
  double trio_score = 0;
  while(it3 != it3_end)
    {
      if(std::binary_search(
           spectrum_code_list.begin(), spectrum_code_list.end(), *it3))
        {
          trio_score += 6;
        }
      else
        {
          trio_score += 0.5;
        }
      it3++;
    }
 
 
  // quatro
  auto it4_end        = it_quatro + 2;
  auto it4            = it_quatro;
  double quatro_score = 0;
  while(it4 != it4_end)
    {
      // element < value
      if(std::binary_search(
           spectrum_code_list.begin(), spectrum_code_list.end(), *it4))
        {
          quatro_score += 8;
        }
      quatro_score += 0.8;
      it4++;
    }
 
  // cinqo
  if(std::binary_search(
       spectrum_code_list.begin(), spectrum_code_list.end(), *it_cinqo))
    {
      score += 10;
    }
  else
    {
      score += 1;
    }
  return score * single_score * duo_score * trio_score * quatro_score;
}

Referenced by convolution().

getPeptideCodedFragment()

const std::vector< std::uint32_t > & pappso::ProteinIntegerCode::getPeptideCodedFragment

(

std::size_t

size

)

const

Definition at line 108 of file proteinintegercode.cpp.

{
  if(size < 2)
    {
 
      throw ExceptionOutOfRange(QObject::tr("size too small"));
    }
  std::size_t indice = size - 2;
  if(indice < m_peptideCodedFragments.size())
    {
      return m_peptideCodedFragments.at(indice);
    }
 
  throw ExceptionOutOfRange(QObject::tr("size too big"));
}

References m_peptideCodedFragments.

Referenced by pappso::ProteinPresenceAbsenceMatrix::fillMatrix().

getProteinSp()

pappso::ProteinSp pappso::ProteinIntegerCode::getProteinSp

(

)

const

Definition at line 80 of file proteinintegercode.cpp.

{
  return msp_protein;
}

References msp_protein.

getSeqAaCode()

const std::vector< std::uint8_t > & pappso::ProteinIntegerCode::getSeqAaCode

(

)

const

Definition at line 74 of file proteinintegercode.cpp.

{
  return m_seqAaCode;
}

References m_seqAaCode.

Referenced by pappso::ProteinPresenceAbsenceMatrix::fillMatrix().

match()

std::vector< std::pair< std::size_t, std::uint32_t > > pappso::ProteinIntegerCode::match

(

const std::vector< uint32_t > &

code_list

)

const

list of positions and matched codes along protein sequence

Definition at line 126 of file proteinintegercode.cpp.

{
  std::vector<std::pair<std::size_t, std::uint32_t>> return_pos;
  std::vector<uint32_t> code_list = code_list_in;
 
  std::sort(code_list.begin(), code_list.end());
  auto it_end = std::unique(code_list.begin(), code_list.end());
  for(auto it_code = code_list.begin(); it_code != it_end; it_code++)
    {
 
      std::size_t size = 2;
      for(auto &liste_protein_seq_code : m_peptideCodedFragments)
        {
 
          auto it_seq_position = std::find(liste_protein_seq_code.begin(),
                                           liste_protein_seq_code.end(),
                                           *it_code);
          while(it_seq_position != liste_protein_seq_code.end())
            {
              // found
              std::size_t position =
                std::distance(liste_protein_seq_code.begin(), it_seq_position);
              return_pos.push_back({size, position});
 
              it_seq_position = std::find(
                ++it_seq_position, liste_protein_seq_code.end(), *it_code);
              qDebug();
            }
          size++;
          qDebug();
        }
      qDebug();
    }
 
  return return_pos;
}

References m_peptideCodedFragments.

Member Data Documentation

m_peptideCodedFragments

std::vector<std::vector<std::uint32_t> > pappso::ProteinIntegerCode::m_peptideCodedFragments

private

Definition at line 104 of file proteinintegercode.h.

Referenced by ProteinIntegerCode(), ProteinIntegerCode(), convolution(), getPeptideCodedFragment(), and match().

m_seqAaCode

std::vector<std::uint8_t> pappso::ProteinIntegerCode::m_seqAaCode

private

Definition at line 103 of file proteinintegercode.h.

Referenced by ProteinIntegerCode(), ProteinIntegerCode(), computePeptideCodeFragments(), convolution(), and getSeqAaCode().

msp_protein

ProteinSp pappso::ProteinIntegerCode::msp_protein

private

Definition at line 101 of file proteinintegercode.h.

Referenced by ProteinIntegerCode(), ProteinIntegerCode(), and getProteinSp().

---

The documentation for this class was generated from the following files:

• pappsomspp/core/protein/proteinintegercode.h
• pappsomspp/core/protein/proteinintegercode.cpp