#include <enzyme.h>

Public Member Functions
	Enzyme ()
	build the default enzyme (trypsin) with recognition_site = "([KR])([^P])"
	Enzyme (const QString &recognition_site)
	build any enzyme given a recognition_site
	~Enzyme ()
void	eat (std::int8_t sequence_database_id, const ProteinSp &protein_sp, bool is_decoy, EnzymeProductInterface &enzyme_product) const
	digest a protein into enzyme products
void	setMiscleavage (unsigned int miscleavage)
	sets the maximum number of missed cleavage allowed in the digestion
unsigned int	getMiscleavage () const
	get the maximum number of missed cleavage allowed in the digestion
void	setTakeOnlyFirstWildcard (bool take_only_first_wildcard)
	take only first m_takeOnlyFirstWildcard
void	setMaxPeptideVariantListSize (std::size_t max_peptide_variant_list_size)
	if there are wildcards in the protein sequence : restrict the number of possible peptide sequences
const QRegularExpression &	getQRegExpRecognitionSite () const

Private Member Functions
void	sanityCheck (EnzymeProductInterface &enzyme_product, std::int8_t sequence_database_id, const ProteinSp &protein_sp, bool is_decoy, const PeptideStr &peptide, unsigned int start, bool is_nter, unsigned int missed_cleavage_number, bool semi_enzyme) const
void	replaceWildcards (std::vector< std::string > *p_peptide_variant_list) const

Private Attributes
QRegularExpression	m_recognitionSite
	example with a kinase == [K,R]
unsigned int	m_miscleavage = 0
bool	m_takeOnlyFirstWildcard = false
std::size_t	m_maxPeptideVariantListSize = 100
std::vector< char >	m_wildCardX
std::vector< char >	m_wildCardB
std::vector< char >	m_wildCardZ

Detailed Description

Definition at line 31 of file enzyme.h.

Constructor & Destructor Documentation

◆ Enzyme() [1/2]

pappso::Enzyme::Enzyme ( )

build the default enzyme (trypsin) with recognition_site = "([KR])([^P])"

Definition at line 32 of file enzyme.cpp.

{
  m_recognitionSite.setPattern("([KR])([^P])");
  m_miscleavage = 0;
 
 
  char vv1[] = {'A', 'R', 'N', 'D', 'C', 'Q', 'E', 'G', 'H', 'I',
                'L', 'K', 'M', 'F', 'P', 'S', 'T', 'W', 'Y', 'V'};
  m_wildCardX.assign(std::begin(vv1), std::end(vv1));
 
  char vv2[] = {'N', 'D'};
  m_wildCardB.assign(std::begin(vv2), std::end(vv2));
 
  char vv3[] = {'Q', 'E'};
  m_wildCardZ.assign(std::begin(vv3), std::end(vv3));
}

References m_miscleavage, m_recognitionSite, m_wildCardB, m_wildCardX, and m_wildCardZ.

◆ Enzyme() [2/2]

pappso::Enzyme::Enzyme ( const QString & recognition_site )

build any enzyme given a recognition_site

Parameters

recognition_site is a regular expression that must identify 2 motifs : one on Nter side one on Cter side

Definition at line 49 of file enzyme.cpp.

{
  m_recognitionSite.setPattern(recognition_site);
  m_miscleavage = 0;
 
 
  char vv1[] = {'A', 'R', 'N', 'D', 'C', 'Q', 'E', 'G', 'H', 'I',
                'L', 'K', 'M', 'F', 'P', 'S', 'T', 'W', 'Y', 'V'};
  m_wildCardX.assign(std::begin(vv1), std::end(vv1));
 
  char vv2[] = {'N', 'D'};
  m_wildCardB.assign(std::begin(vv2), std::end(vv2));
 
  char vv3[] = {'Q', 'E'};
  m_wildCardZ.assign(std::begin(vv3), std::end(vv3));
}

References m_miscleavage, m_recognitionSite, m_wildCardB, m_wildCardX, and m_wildCardZ.

◆ ~Enzyme()

pappso::Enzyme::~Enzyme ( )

Definition at line 66 of file enzyme.cpp.

67{

68}

Member Function Documentation

◆ eat()

void pappso::Enzyme::eat	(	std::int8_t	sequence_database_id,
		const ProteinSp &	protein_sp,
		bool	is_decoy,
		EnzymeProductInterface &	enzyme_product ) const

digest a protein into enzyme products

Parameters

sequence_database_id	integer that references the sequence fatabase (file, stream, url...)
protein_sp	is the original protein to be digested
is_decoy	tell if the current protein is a decoy (true) or normal (false) protein
enzyme_product	is the object that will receive the digestion products

Definition at line 87 of file enzyme.cpp.

{
  /*
   *        for aa in self.aa_to_cut:
            seq = seq.replace(aa, aa + ' ')
        seq_stack = []
        for s in seq.strip().split(' '):
            seq_stack.append(s)
            if len(seq_stack) > self.misscleavage + 1:
                seq_stack.pop(0)
            s2 = ""
            for s_miss in seq_stack[::-1]:
                s2 = s_miss + s2
                yield s2
  */
  qDebug() << "Enzyme::eat begin ";
  const QString sequence = protein_sp.get()->getSequence();
  qDebug() << sequence;
  QStringList peptide_list;
  int pos                                        = 0;
  int peptide_start                              = 0;
  int peptide_size                               = sequence.size();
  QRegularExpressionMatch match_recognition_site = m_recognitionSite.match(sequence, pos);
  while(match_recognition_site.hasMatch())
    {
      pos          = match_recognition_site.capturedStart(0);
      peptide_size = pos + match_recognition_site.captured(1).length() - peptide_start;
      // qDebug() << "pos=" << pos << " peptide_start=" << peptide_start << "
      // peptide_size=" << peptide_size << " " <<
      // sequence.mid(peptide_start,peptide_size);
      if(peptide_size > 0)
        {
          peptide_list.append(sequence.mid(peptide_start, peptide_size));
        }
      peptide_start += peptide_size;
      pos                    = peptide_start; // all peptides MUST be consecutive
      match_recognition_site = m_recognitionSite.match(sequence, pos);
    }
  peptide_size = sequence.size() - peptide_start;
  if(peptide_size > 0)
    {
      peptide_list.append(sequence.mid(peptide_start, peptide_size));
    }
 
  unsigned int start = 1;
  bool is_nter       = true;
  foreach(const QString &peptide, peptide_list)
    {
      // enzyme_product.setPeptide(sequence_database_id, protein_sp,is_decoy,
      // peptide, start,is_nter,0, false);
      sanityCheck(enzyme_product,
                  sequence_database_id,
                  protein_sp,
                  is_decoy,
                  peptide,
                  start,
                  is_nter,
                  0,
                  false);
      is_nter = false;
      start += peptide.size();
    }
 
  unsigned int miscleavage_i = 0;
  while(miscleavage_i < m_miscleavage)
    {
      miscleavage_i++;
      qDebug() << "miscleavage_i=" << miscleavage_i;
      int chunk_number   = miscleavage_i + 1;
      unsigned int start = 1;
      bool is_nter       = true;
 
      for(auto i = 0; i < peptide_list.size(); ++i)
        {
          qDebug() << "start=" << start;
          QStringList peptide_mis_list;
          for(auto j = 0; (j < chunk_number) && ((i + j) < peptide_list.size()); j++)
            {
              peptide_mis_list << peptide_list.at(i + j);
            }
          if(peptide_mis_list.size() == chunk_number)
            {
              // enzyme_product.setPeptide(sequence_database_id,
              // protein_sp,is_decoy, peptide_mis_list.join(""), start,is_nter,
              // miscleavage_i, false);
              sanityCheck(enzyme_product,
                          sequence_database_id,
                          protein_sp,
                          is_decoy,
                          peptide_mis_list.join(""),
                          start,
                          is_nter,
                          miscleavage_i,
                          false);
            }
          is_nter = false;
          start += peptide_list.at(i).size();
        }
    }
}

References m_miscleavage, m_recognitionSite, and sanityCheck().

◆ getMiscleavage()

unsigned int pappso::Enzyme::getMiscleavage ( ) const

get the maximum number of missed cleavage allowed in the digestion

Returns: miscleavage maximum number of missed cleavade to allow (defaults is 0)

Definition at line 76 of file enzyme.cpp.

{
  return m_miscleavage;
}

References m_miscleavage.

◆ getQRegExpRecognitionSite()

const QRegularExpression & pappso::Enzyme::getQRegExpRecognitionSite ( ) const

Definition at line 348 of file enzyme.cpp.

{
  return m_recognitionSite;
}

References m_recognitionSite.

◆ replaceWildcards()

void pappso::Enzyme::replaceWildcards ( std::vector< std::string > * p_peptide_variant_list ) const

private

Definition at line 192 of file enzyme.cpp.

{
  std::string new_peptide = p_peptide_variant_list->at(0);
  qDebug() << "Enzyme::replaceWildcards begin " << new_peptide.c_str();
  std::vector<std::string> old_peptide_variant_list;
  old_peptide_variant_list.assign(p_peptide_variant_list->begin(), p_peptide_variant_list->end());
 
 
  for(char wildcard : {'X', 'B', 'Z'})
    {
 
      std::size_t position = new_peptide.find(wildcard);
      if(position == std::string::npos)
        {
          continue;
        }
      else
        {
          p_peptide_variant_list->clear();
          /*
          new_peptide[position] = 'A';
          p_peptide_variant_list->push_back(new_peptide);
          break;
          */
 
          const std::vector<char> *p_x_replace_wildcard = nullptr;
          if(wildcard == 'X')
            {
              p_x_replace_wildcard = &m_wildCardX;
            }
          else if(wildcard == 'B')
            {
              p_x_replace_wildcard = &m_wildCardB;
            }
          else if(wildcard == 'Z')
            {
              p_x_replace_wildcard = &m_wildCardZ;
            }
 
          if(p_x_replace_wildcard != nullptr)
            {
              for(std::string orig_peptide : old_peptide_variant_list)
                {
                  for(char replace : *p_x_replace_wildcard)
                    {
                      orig_peptide[position] = replace;
                      p_peptide_variant_list->push_back(orig_peptide);
                    }
                }
            }
          else
            {
              throw ExceptionNotPossible(QObject::tr("x_replace_wildcard is empty"));
            }
          // new_peptide[position] = 'A';
          // p_peptide_variant_list->push_back(new_peptide);
          // p_peptide_variant_list->resize(1);
          // std::cerr << "Enzyme::replaceWildcards begin
          // p_peptide_variant_list.size()=" << p_peptide_variant_list->size()
          // <<
          // endl;
          break;
        }
    }
  std::vector<std::string>().swap(
    old_peptide_variant_list); // clear old_peptide_variant_list reallocating
 
 
  qDebug() << "Enzyme::replaceWildcards end " << new_peptide.c_str();
}

References m_wildCardB, m_wildCardX, and m_wildCardZ.

Referenced by sanityCheck().

◆ sanityCheck()

void pappso::Enzyme::sanityCheck	(	EnzymeProductInterface &	enzyme_product,
		std::int8_t	sequence_database_id,
		const ProteinSp &	protein_sp,
		bool	is_decoy,
		const PeptideStr &	peptide,
		unsigned int	start,
		bool	is_nter,
		unsigned int	missed_cleavage_number,
		bool	semi_enzyme ) const

private

Definition at line 271 of file enzyme.cpp.

{
  if(peptide.contains('X') || peptide.contains('B') || peptide.contains('Z'))
    {
 
      std::vector<std::string> peptide_variant_list;
      peptide_variant_list.push_back(peptide.toStdString());
 
      while((peptide_variant_list.at(0).find('X') != std::string::npos) ||
            (peptide_variant_list.at(0).find('B') != std::string::npos) ||
            (peptide_variant_list.at(0).find('Z') != std::string::npos))
        {
          replaceWildcards(&peptide_variant_list);
          if(peptide_variant_list.size() > m_maxPeptideVariantListSize)
            {
              peptide_variant_list.resize(m_maxPeptideVariantListSize);
              peptide_variant_list.shrink_to_fit();
            }
        }
 
      // peptide_variant_list.resize(2);
      if(m_takeOnlyFirstWildcard)
        {
          enzyme_product.setPeptide(sequence_database_id,
                                    protein_sp,
                                    is_decoy,
                                    QString(peptide_variant_list.at(0).c_str()),
                                    start,
                                    is_nter,
                                    missed_cleavage_number,
                                    semi_enzyme);
        }
      else
        {
          std::string peptide_variant = peptide_variant_list.back();
          while(peptide_variant_list.size() > 0)
            {
              enzyme_product.setPeptide(sequence_database_id,
                                        protein_sp,
                                        is_decoy,
                                        QString(peptide_variant.c_str()),
                                        start,
                                        is_nter,
                                        missed_cleavage_number,
                                        semi_enzyme);
              peptide_variant_list.pop_back();
              if(peptide_variant_list.size() > 0)
                {
                  peptide_variant = peptide_variant_list.back();
                }
            }
        }
      std::vector<std::string>().swap(
        peptide_variant_list); // clear peptide_variant_list reallocating
    }
  else
    {
      enzyme_product.setPeptide(sequence_database_id,
                                protein_sp,
                                is_decoy,
                                peptide,
                                start,
                                is_nter,
                                missed_cleavage_number,
                                semi_enzyme);
    }
}

References m_maxPeptideVariantListSize, m_takeOnlyFirstWildcard, replaceWildcards(), and pappso::EnzymeProductInterface::setPeptide().

Referenced by eat().

◆ setMaxPeptideVariantListSize()

void pappso::Enzyme::setMaxPeptideVariantListSize ( std::size_t max_peptide_variant_list_size )

if there are wildcards in the protein sequence : restrict the number of possible peptide sequences

Parameters

max_peptide_variant_list_size maximum number of peptide variant (default is 100)

Definition at line 81 of file enzyme.cpp.

{
  m_maxPeptideVariantListSize = max_peptide_variant_list_size;
}

References m_maxPeptideVariantListSize.

◆ setMiscleavage()

void pappso::Enzyme::setMiscleavage ( unsigned int miscleavage )

sets the maximum number of missed cleavage allowed in the digestion

Parameters

miscleavage maximum number of missed cleavade to allow (defaults is 0)

Definition at line 71 of file enzyme.cpp.

{
  m_miscleavage = miscleavage;
}

References m_miscleavage.

◆ setTakeOnlyFirstWildcard()

void pappso::Enzyme::setTakeOnlyFirstWildcard ( bool take_only_first_wildcard )

take only first m_takeOnlyFirstWildcard

Parameters

bool	true : switch to take only the first possibility if there are X, B or Z wildcards in sequence

Definition at line 264 of file enzyme.cpp.

{
  m_takeOnlyFirstWildcard = take_only_first_wildcard;
}

References m_takeOnlyFirstWildcard.

Member Data Documentation

◆ m_maxPeptideVariantListSize

std::size_t pappso::Enzyme::m_maxPeptideVariantListSize = 100

private

Definition at line 93 of file enzyme.h.

Referenced by sanityCheck(), and setMaxPeptideVariantListSize().

◆ m_miscleavage

unsigned int pappso::Enzyme::m_miscleavage = 0

private

Definition at line 90 of file enzyme.h.

Referenced by Enzyme(), Enzyme(), eat(), getMiscleavage(), and setMiscleavage().

◆ m_recognitionSite

QRegularExpression pappso::Enzyme::m_recognitionSite

private

example with a kinase == [K,R]

Definition at line 89 of file enzyme.h.

Referenced by Enzyme(), Enzyme(), eat(), and getQRegExpRecognitionSite().

◆ m_takeOnlyFirstWildcard

bool pappso::Enzyme::m_takeOnlyFirstWildcard = false

private

Definition at line 91 of file enzyme.h.

Referenced by sanityCheck(), and setTakeOnlyFirstWildcard().

◆ m_wildCardB

std::vector<char> pappso::Enzyme::m_wildCardB

private

Definition at line 97 of file enzyme.h.

Referenced by Enzyme(), Enzyme(), and replaceWildcards().

◆ m_wildCardX

std::vector<char> pappso::Enzyme::m_wildCardX

private

Definition at line 96 of file enzyme.h.

Referenced by Enzyme(), Enzyme(), and replaceWildcards().

◆ m_wildCardZ

std::vector<char> pappso::Enzyme::m_wildCardZ

private

Definition at line 98 of file enzyme.h.

Referenced by Enzyme(), Enzyme(), and replaceWildcards().

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

pappsomspp/core/protein/enzyme.h
pappsomspp/core/protein/enzyme.cpp

Public Member Functions

Private Member Functions

Private Attributes

Detailed Description

Constructor & Destructor Documentation

◆ Enzyme() [1/2]

◆ Enzyme() [2/2]

◆ ~Enzyme()

Member Function Documentation

◆ eat()

◆ getMiscleavage()

◆ getQRegExpRecognitionSite()

◆ replaceWildcards()

◆ sanityCheck()

◆ setMaxPeptideVariantListSize()

◆ setMiscleavage()

◆ setTakeOnlyFirstWildcard()

Member Data Documentation

◆ m_maxPeptideVariantListSize

◆ m_miscleavage

◆ m_recognitionSite

◆ m_takeOnlyFirstWildcard

◆ m_wildCardB

◆ m_wildCardX

◆ m_wildCardZ