void DetectabilitySimulation::svmFilter_(SimTypes::FeatureMapSim& features)
{
// transform featuremap to peptides vector
vector<String> peptides_vector(features.size());
for (Size i = 0; i < features.size(); ++i)
{
peptides_vector[i] = features[i].getPeptideIdentifications()[0].getHits()[0].getSequence().toUnmodifiedString();
}
vector<double> labels;
vector<double> detectabilities;
predictDetectabilities(peptides_vector, labels, detectabilities);
// copy all meta data stored in the feature map
SimTypes::FeatureMapSim temp_copy(features);
temp_copy.clear(false);
for (Size i = 0; i < peptides_vector.size(); ++i)
{
if (detectabilities[i] > min_detect_)
{
features[i].setMetaValue("detectability", detectabilities[i]);
temp_copy.push_back(features[i]);
}
#ifdef DEBUG_SIM
cout << detectabilities[i] << " " << min_detect_ << endl;
#endif
}
features.swap(temp_copy);
}
void DetectabilitySimulation::noFilter_(SimTypes::FeatureMapSim& features)
{
// set detectibility to 1.0 for all given peptides
double defaultDetectibility = 1.0;
for (SimTypes::FeatureMapSim::iterator feature_it = features.begin();
feature_it != features.end();
++feature_it)
{
(*feature_it).setMetaValue("detectability", defaultDetectibility);
}
}
void ICPLLabeler::addLabelToProteinHits_(SimTypes::FeatureMapSim& features, const String& label) const
{
// check if proteinIdentification exists before accessing it
if (features.getProteinIdentifications().empty())
return;
for (std::vector<ProteinHit>::iterator protein_hit = features.getProteinIdentifications()[0].getHits().begin();
protein_hit != features.getProteinIdentifications()[0].getHits().end();
++protein_hit)
{
AASequence aa = AASequence::fromString(protein_hit->getSequence());
// modify only if the term is accessible
if (aa.getNTerminalModification() == "")
{
aa.setNTerminalModification(label);
protein_hit->setSequence(aa.toString());
}
}
}
void MSSim::createFeatureMap_(const SimTypes::SampleProteins& proteins, SimTypes::FeatureMapSim& feature_map, Size map_index)
{
// clear feature map
feature_map.clear(true);
ProteinIdentification protIdent;
for (SimTypes::SampleProteins::const_iterator it = proteins.begin(); it != proteins.end(); ++it)
{
// add new ProteinHit to ProteinIdentification
ProteinHit protHit(0.0, 1, (it->entry).identifier, (it->entry).sequence);
// copy all meta values from FASTA file parsing
protHit = (it->meta);
// additional meta values:
protHit.setMetaValue("description", it->entry.description);
protHit.setMetaValue("map_index", map_index);
protIdent.insertHit(protHit);
}
vector<ProteinIdentification> vec_protIdent;
vec_protIdent.push_back(protIdent);
feature_map.setProteinIdentifications(vec_protIdent);
}
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());
}
}
void ITRAQLabeler::labelPeptide_(const Feature& feature, SimTypes::FeatureMapSim& result) const
{
// modify with iTRAQ modification (needed for mass calculation and MS/MS signal)
//site="Y" - low abundance
//site="N-term"
//site="K" - lysine
String modification = (itraq_type_ == ItraqConstants::FOURPLEX ? "iTRAQ4plex" : "iTRAQ8plex");
vector<PeptideHit> pep_hits(feature.getPeptideIdentifications()[0].getHits());
AASequence seq(pep_hits[0].getSequence());
// N-term
seq.setNTerminalModification(modification);
// all "K":
for (Size i = 0; i < seq.size(); ++i)
{
if (seq[i] == 'K' && !seq[i].isModified())
seq.setModification(i, modification);
}
result.resize(1);
result[0] = feature;
pep_hits[0].setSequence(seq);
result[0].getPeptideIdentifications()[0].setHits(pep_hits);
// some "Y":
// for each "Y" create two new features, depending on labeling efficiency on "Y":
if (y_labeling_efficiency_ == 0)
return;
for (Size i = 0; i < seq.size(); ++i)
{
if (seq[i] == 'Y' && !seq[i].isModified())
{
if (y_labeling_efficiency_ == 1)
{
addModificationToPeptideHit_(result.back(), modification, i);
}
else // double number of features:
{
Size f_count = result.size();
for (Size f = 0; f < f_count; ++f)
{
// copy feature
result.push_back(result[f]);
// modify the copy
addModificationToPeptideHit_(result.back(), modification, i);
// adjust intensities:
result.back().setIntensity(result.back().getIntensity() * y_labeling_efficiency_);
result[f].setIntensity(result[f].getIntensity() * (1 - y_labeling_efficiency_));
}
}
}
}
}
void DigestSimulation::digest(SimTypes::FeatureMapSim& feature_map)
{
LOG_INFO << "Digest Simulation ... started" << std::endl;
if ((String)param_.getValue("enzyme") == String("none"))
{
//peptides = proteins;
// convert all proteins into peptides
// for each protein_hit in the FeatureMap
for (std::vector<ProteinHit>::iterator protein_hit = feature_map.getProteinIdentifications()[0].getHits().begin();
protein_hit != feature_map.getProteinIdentifications()[0].getHits().end();
++protein_hit)
{
// generate a PeptideHit hit with the correct link to the protein
PeptideHit pep_hit(1.0, 1, 0, AASequence::fromString(protein_hit->getSequence()));
PeptideEvidence pe;
pe.setProteinAccession(protein_hit->getAccession());
pep_hit.addPeptideEvidence(pe);
// add the PeptideHit to the PeptideIdentification
PeptideIdentification pep_id;
pep_id.insertHit(pep_hit);
// generate Feature with correct Intensity and corresponding PeptideIdentification
Feature f;
f.getPeptideIdentifications().push_back(pep_id);
f.setIntensity(protein_hit->getMetaValue("intensity"));
// copy intensity meta-values and additional annotations from Protein to Feature
StringList keys;
protein_hit->getKeys(keys);
for (StringList::const_iterator it_key = keys.begin(); it_key != keys.end(); ++it_key)
{
f.setMetaValue(*it_key, protein_hit->getMetaValue(*it_key));
}
// add Feature to SimTypes::FeatureMapSim
feature_map.push_back(f);
}
return;
}
UInt min_peptide_length = param_.getValue("min_peptide_length");
bool use_log_model = param_.getValue("model") == "trained" ? true : false;
UInt missed_cleavages = param_.getValue("model_naive:missed_cleavages");
double cleave_threshold = param_.getValue("model_trained:threshold");
EnzymaticDigestion digestion;
digestion.setEnzyme(digestion.getEnzymeByName((String)param_.getValue("enzyme")));
digestion.setLogModelEnabled(use_log_model);
digestion.setLogThreshold(cleave_threshold);
std::vector<AASequence> digestion_products;
// keep track of generated features
std::map<AASequence, Feature> generated_features;
// Iterate through ProteinHits in the FeatureMap and digest them
for (std::vector<ProteinHit>::iterator protein_hit = feature_map.getProteinIdentifications()[0].getHits().begin();
protein_hit != feature_map.getProteinIdentifications()[0].getHits().end();
++protein_hit)
{
// determine abundance of each digestion product (this is quite long now...)
// we assume that each digestion product will have the same abundance
// note: missed cleavages reduce overall abundance as they combine two (or more) single peptides
// how many "atomic"(i.e. non-cleavable) peptides are created?
digestion.setMissedCleavages(0);
Size complete_digest_count = digestion.peptideCount(AASequence::fromString(protein_hit->getSequence()));
// compute average number of "atomic" peptides summed from all digestion products
Size number_atomic_whole = 0;
Size number_of_digestion_products = 0;
for (Size i = 0; (i <= missed_cleavages) && (i < complete_digest_count); ++i)
{
number_atomic_whole += (complete_digest_count - i) * (i + 1);
number_of_digestion_products += (complete_digest_count - i);
}
// mean number of "atomic" peptides per digestion product is now: number_atomic_whole / number_of_digestion_products
// -> thus abundance of a digestion product is: #proteins / avg#of"atomic"peptides
// i.e.: protein->second / (number_atomic_whole / number_of_digestion_products)
Map<String, SimTypes::SimIntensityType> intensities;
StringList keys;
protein_hit->getKeys(keys);
for (StringList::const_iterator it_key = keys.begin(); it_key != keys.end(); ++it_key)
{
if (!it_key->hasPrefix("intensity"))
continue;
intensities[*it_key] = std::max(SimTypes::SimIntensityType(1), SimTypes::SimIntensityType(protein_hit->getMetaValue(*it_key))
* SimTypes::SimIntensityType(number_of_digestion_products)
/ SimTypes::SimIntensityType(number_atomic_whole)); // order changed for numeric stability
}
// do real digest
digestion.setMissedCleavages(missed_cleavages);
digestion.digest(AASequence::fromString(protein_hit->getSequence()), digestion_products);
for (std::vector<AASequence>::const_iterator dp_it = digestion_products.begin();
dp_it != digestion_products.end();
++dp_it)
{
if (dp_it->size() < min_peptide_length)
continue;
// sum equal peptide's intensities
// *dp_it -> peptide
// If we see this Peptide the first time -> generate corresponding feature
if (generated_features.count(*dp_it) == 0)
{
PeptideHit pep_hit(1.0, 1, 0, *dp_it);
PeptideIdentification pep_id;
pep_id.insertHit(pep_hit);
// create feature
Feature f;
f.getPeptideIdentifications().push_back(pep_id);
// set intensity to 0 to avoid problems when summing up
f.setIntensity(0.0);
// copy all non-intensity meta values
StringList lkeys;
protein_hit->getKeys(lkeys);
for (StringList::iterator key = lkeys.begin(); key != lkeys.end(); ++key)
{
if (!key->hasPrefix("intensity"))
{
f.setMetaValue(*key, protein_hit->getMetaValue(*key));
}
}
// insert into map
generated_features.insert(std::make_pair(*dp_it, f));
}
// sum up intensity values
generated_features[*dp_it].setIntensity(generated_features[*dp_it].getIntensity() + intensities["intensity"]);
// ... same for other intensities (iTRAQ...)
for (Map<String, SimTypes::SimIntensityType>::const_iterator it_other = intensities.begin(); it_other != intensities.end(); ++it_other)
{
if (!generated_features[*dp_it].metaValueExists(it_other->first))
{
generated_features[*dp_it].setMetaValue(it_other->first, it_other->second);
}
else
{
generated_features[*dp_it].setMetaValue(it_other->first, SimTypes::SimIntensityType(generated_features[*dp_it].getMetaValue(it_other->first)) + it_other->second);
}
}
// add current protein accession
// existing proteins accessions ...
std::set<String> protein_accessions = generated_features[*dp_it].getPeptideIdentifications()[0].getHits()[0].extractProteinAccessions();
// ... add accession of current protein
protein_accessions.insert(protein_hit->getAccession());
std::vector<PeptideIdentification> pep_idents = generated_features[*dp_it].getPeptideIdentifications();
std::vector<PeptideHit> pep_hits = pep_idents[0].getHits();
for (std::set<String>::const_iterator s_it = protein_accessions.begin(); s_it != protein_accessions.end(); ++s_it)
{
PeptideEvidence pe;
pe.setProteinAccession(*s_it);
pep_hits[0].addPeptideEvidence(pe);
}
pep_idents[0].setHits(pep_hits);
generated_features[*dp_it].setPeptideIdentifications(pep_idents);
}
}
// add generated_features to FeatureMap
for (std::map<AASequence, Feature>::iterator it_gf = generated_features.begin();
it_gf != generated_features.end();
++it_gf)
{
// round up intensity
(it_gf->second).setIntensity(ceil((it_gf->second).getIntensity()));
feature_map.push_back(it_gf->second);
}
}
