void EnzymaticDigestionLogModel::digest(const AASequence& protein, vector<AASequence>& output) const
{
// initialization
output.clear();
AASequence::ConstIterator begin = protein.begin();
AASequence::ConstIterator end = protein.begin();
while (nextCleavageSite_(protein, end), end != protein.end())
{
output.push_back(protein.getSubsequence(begin - protein.begin(), end - begin));
begin = end;
}
output.push_back(protein.getSubsequence(begin - protein.begin(), end - begin));
}
bool EnzymaticDigestionLogModel::isCleavageSite_(
const AASequence& protein, const AASequence::ConstIterator& iterator) const
{
if (enzyme_.getName() != "Trypsin") // no cleavage
{
throw Exception::InvalidParameter(__FILE__, __LINE__, __PRETTY_FUNCTION__, String("EnzymaticDigestionLogModel: enzyme '") + enzyme_.getName() + " does not support logModel!");
}
else
{
if ((!enzyme_.getRegEx().hasSubstring(iterator->getOneLetterCode())) || *iterator == 'P') // wait for R or K
{
return false;
}
SignedSize pos = distance(AASequence::ConstIterator(protein.begin()),
iterator) - 4; // start position in sequence
double score_cleave = 0, score_missed = 0;
for (SignedSize i = 0; i < 9; ++i)
{
if ((pos + i >= 0) && (pos + i < (SignedSize)protein.size()))
{
BindingSite_ bs(i, protein[pos + i].getOneLetterCode());
Map<BindingSite_, CleavageModel_>::const_iterator pos_it =
model_data_.find(bs);
if (pos_it != model_data_.end()) // no data for non-std. amino acids
{
score_cleave += pos_it->second.p_cleave;
score_missed += pos_it->second.p_miss;
}
}
}
return score_missed - score_cleave > log_model_threshold_;
}
}
Size EnzymaticDigestionLogModel::peptideCount(const AASequence& protein)
{
SignedSize count = 1;
AASequence::ConstIterator iterator = protein.begin();
while (nextCleavageSite_(protein, iterator), iterator != protein.end())
{
++count;
}
Size sum = count;
return sum;
}
void SILACLabeler::applyLabelToProteinHit_(SimTypes::FeatureMapSim& channel, const String& arginine_label, const String& lysine_label) const
{
for (std::vector<ProteinHit>::iterator protein_hit = channel.getProteinIdentifications()[0].getHits().begin();
protein_hit != channel.getProteinIdentifications()[0].getHits().end();
++protein_hit)
{
AASequence aa = AASequence::fromString(protein_hit->getSequence());
for (AASequence::Iterator residue = aa.begin(); residue != aa.end(); ++residue)
{
if (*residue == 'R')
{
aa.setModification(residue - aa.begin(), arginine_label);
}
else if (*residue == 'K')
{
aa.setModification(residue - aa.begin(), lysine_label);
}
}
protein_hit->setSequence(aa.toString());
}
}
String CompNovoIdentificationBase::getModifiedStringFromAASequence_(const AASequence & sequence)
{
String seq;
for (AASequence::ConstIterator it = sequence.begin(); it != sequence.end(); ++it)
{
if (residue_to_name_.has(&*it))
{
seq += residue_to_name_[&*it];
}
else
{
seq += it->getOneLetterCode();
}
}
return seq;
}
String IBSpectraFile::getModifString_(const AASequence& sequence)
{
String modif = sequence.getNTerminalModification();
for (AASequence::ConstIterator aa_it = sequence.begin();
aa_it != sequence.end();
++aa_it)
{
modif += ":" + aa_it->getModification();
}
if (sequence.getCTerminalModification() != "")
{
modif += ":" + sequence.getCTerminalModification();
}
return modif;
}
void TheoreticalSpectrumGenerator::addLosses_(RichPeakSpectrum & spectrum, const AASequence & ion, double intensity, Residue::ResidueType res_type, int charge) const
{
RichPeak1D p;
set<String> losses;
for (AASequence::ConstIterator it = ion.begin(); it != ion.end(); ++it)
{
if (it->hasNeutralLoss())
{
vector<EmpiricalFormula> loss_formulas = it->getLossFormulas();
for (Size i = 0; i != loss_formulas.size(); ++i)
{
losses.insert(loss_formulas[i].toString());
}
}
}
if (!add_isotopes_)
{
p.setIntensity(intensity * rel_loss_intensity_);
}
for (set<String>::const_iterator it = losses.begin(); it != losses.end(); ++it)
{
EmpiricalFormula loss_ion = ion.getFormula(res_type, charge) - EmpiricalFormula(*it);
// thanks to Chris and Sandro
// check for negative element frequencies (might happen if losses are not allowed for specific ions)
bool negative_elements(false);
for (EmpiricalFormula::ConstIterator eit = loss_ion.begin(); eit != loss_ion.end(); ++eit)
{
if (eit->second < 0)
{
negative_elements = true;
break;
}
}
if (negative_elements)
{
continue;
}
double loss_pos = loss_ion.getMonoWeight() / (double)charge;
const String& loss_name = *it;
if (add_isotopes_)
{
IsotopeDistribution dist = loss_ion.getIsotopeDistribution(max_isotope_);
UInt j(0);
for (IsotopeDistribution::ConstIterator iso = dist.begin(); iso != dist.end(); ++iso)
{
p.setMZ((double)(loss_pos + j) / (double)charge);
p.setIntensity(intensity * rel_loss_intensity_ * iso->second);
if (add_metainfo_ && j == 0)
{
// note: important to construct a string from char. If omitted it will perform pointer arithmetics on the "-" string literal
String ion_name = String(residueTypeToIonLetter_(res_type)) + String(ion.size()) + "-" + loss_name + String(charge, '+');
p.setMetaValue("IonName", ion_name);
}
spectrum.push_back(p);
}
}
else
{
p.setMZ(loss_pos);
if (add_metainfo_)
{
// note: important to construct a string from char. If omitted it will perform pointer arithmetics on the "-" string literal
String ion_name = String(residueTypeToIonLetter_(res_type)) + String(ion.size()) + "-" + loss_name + String(charge, '+');
p.setMetaValue("IonName", ion_name);
}
spectrum.push_back(p);
}
}
}
bool ModificationDefinitionsSet::isCompatible(const AASequence & peptide) const
{
set<String> var_names(getVariableModificationNames()), fixed_names(getFixedModificationNames());
// no modifications present and needed
if (fixed_names.empty() && !peptide.isModified())
{
return true;
}
// check whether the fixed modifications are fulfilled
if (!fixed_names.empty())
{
for (set<String>::const_iterator it1 = fixed_names.begin(); it1 != fixed_names.end(); ++it1)
{
String origin = ModificationsDB::getInstance()->getModification(*it1).getOrigin();
// only single 1lc amino acids are allowed
if (origin.size() != 1)
{
continue;
}
for (AASequence::ConstIterator it2 = peptide.begin(); it2 != peptide.end(); ++it2)
{
if (origin == it2->getOneLetterCode())
{
// check whether the residue is modified (has to be)
if (!it2->isModified())
{
return false;
}
// check whether the modification is the same
if (ModificationsDB::getInstance()->getModification(*it1).getId() != it2->getModification())
{
return false;
}
}
}
}
}
// check wether other modifications than the variable are present
for (AASequence::ConstIterator it = peptide.begin(); it != peptide.end(); ++it)
{
if (it->isModified())
{
String mod = ModificationsDB::getInstance()->getModification(it->getOneLetterCode(), it->getModification(), ResidueModification::ANYWHERE).getFullId();
if (var_names.find(mod) == var_names.end() &&
fixed_names.find(mod) == fixed_names.end())
{
return false;
}
}
}
if (peptide.hasNTerminalModification())
{
String mod = ModificationsDB::getInstance()->getTerminalModification(peptide.getNTerminalModification(), ResidueModification::N_TERM).getFullId();
if (var_names.find(mod) == var_names.end() &&
fixed_names.find(mod) == fixed_names.end())
{
return false;
}
}
if (peptide.hasCTerminalModification())
{
String mod = ModificationsDB::getInstance()->getTerminalModification(peptide.getCTerminalModification(), ResidueModification::C_TERM).getFullId();
if (var_names.find(mod) == var_names.end() &&
fixed_names.find(mod) == fixed_names.end())
{
return false;
}
}
return true;
}