// merge all channels into the first one
// no further influence of the simulation process needed
void LabelFreeLabeler::setUpHook(SimTypes::FeatureMapSimVector& features)
{
if (features.size() == 1)
return;
else
{
LOG_INFO << "Merging input FASTA files into one. Intensities will be summed up if duplicates occur.";
SimTypes::FeatureMapSim final_map = mergeProteinIdentificationsMaps_(features);
features.clear();
features.push_back(final_map);
}
}
void digestFeaturesMapSimVector_(SimTypes::FeatureMapSimVector& feature_maps)
{
// digest here
DigestSimulation digest_sim;
Param p;
p.setValue("model", "naive");
p.setValue("model_naive:missed_cleavages", 0);
digest_sim.setParameters(p);
std::cout << digest_sim.getParameters() << std::endl;
for(SimTypes::FeatureMapSimVector::iterator iter = feature_maps.begin() ; iter != feature_maps.end() ; ++iter)
{
digest_sim.digest((*iter));
}
}
/// Labeling between digestion and rt simulation
/// Join all peptides with the same sequence into one feature
/// channels are retained via metavalues
/// if a peptide is not present in all channels, then there will be missing meta values! (so don't rely on them being present)
void ITRAQLabeler::postDigestHook(SimTypes::FeatureMapSimVector& channels)
{
// merge channels into a single feature map
SimTypes::FeatureMapSim final_feature_map = mergeProteinIdentificationsMaps_(channels);
std::map<String, Size> peptide_to_feature;
for (Size i = 0; i < channels.size(); ++i)
{
for (SimTypes::FeatureMapSim::iterator it_f_o = channels[i].begin();
it_f_o != channels[i].end();
++it_f_o)
{
// derive iTRAQ labeled features from original sequence (might be more than one due to partial labeling)
SimTypes::FeatureMapSim labeled_features;
labelPeptide_(*it_f_o, labeled_features);
for (SimTypes::FeatureMapSim::iterator it_f = labeled_features.begin();
it_f != labeled_features.end();
++it_f)
{
const String& seq = it_f->getPeptideIdentifications()[0].getHits()[0].getSequence().toString();
Size f_index;
//check if we already have a feature for this peptide
if (peptide_to_feature.count(seq) > 0)
{
f_index = peptide_to_feature[seq];
}
else // create new feature
{
final_feature_map.push_back(*it_f);
// update map:
f_index = final_feature_map.size() - 1;
peptide_to_feature[seq] = f_index;
}
// add intensity as metavalue
final_feature_map[f_index].setMetaValue(getChannelIntensityName(i), it_f->getIntensity());
// increase overall intensity
final_feature_map[f_index].setIntensity(final_feature_map[f_index].getIntensity() + it_f->getIntensity());
mergeProteinAccessions_(final_feature_map[f_index], *it_f);
}
}
}
channels.clear();
channels.push_back(final_feature_map);
}
void ITRAQLabeler::postRawTandemMSHook(SimTypes::FeatureMapSimVector& fm, SimTypes::MSSimExperiment& exp)
{
//std::cout << "Matrix used: \n" << ItraqConstants::translateIsotopeMatrix(itraq_type_, isotope_corrections_) << "\n\n";
double rep_shift = param_.getValue("reporter_mass_shift");
OPENMS_PRECONDITION(fm.size() == 1, "More than one feature map given in ITRAQLabeler::postRawTandemMSHook()!")
EigenMatrixXdPtr channel_frequency = convertOpenMSMatrix2EigenMatrixXd(ItraqConstants::translateIsotopeMatrix(itraq_type_, isotope_corrections_));
Eigen::MatrixXd itraq_intensity_sum(ItraqConstants::CHANNEL_COUNT[itraq_type_], 1);
std::vector<Matrix<Int> > channel_names(2);
channel_names[0].setMatrix<4, 1>(ItraqConstants::CHANNELS_FOURPLEX);
channel_names[1].setMatrix<8, 1>(ItraqConstants::CHANNELS_EIGHTPLEX);
boost::uniform_real<double> udist(0.0, 1.0);
// add signal...
for (SimTypes::MSSimExperiment::iterator it = exp.begin(); it != exp.end(); ++it)
{
if (it->getMSLevel() != 2)
continue;
// reset sum matrix to 0
itraq_intensity_sum.setZero();
// add up signal of all features
OPENMS_PRECONDITION(it->metaValueExists("parent_feature_ids"), "Meta value 'parent_feature_ids' missing in ITRAQLabeler::postRawTandemMSHook()!")
IntList parent_fs = it->getMetaValue("parent_feature_ids");
for (Size i_f = 0; i_f < parent_fs.size(); ++i_f)
{
// get RT scaled iTRAQ intensities
EigenMatrixXdPtr row = getItraqIntensity_(fm[0][parent_fs[i_f]], it->getRT());
// apply isotope matrix to active channels
// row * channel_frequencyOld = observed iTRAQ intensities
Eigen::MatrixXd itraq_intensity_observed = (*channel_frequency) * (*row);
// add result to sum
itraq_intensity_sum += itraq_intensity_observed;
}
// add signal to MS2 spectrum
for (Int i_channel = 0; i_channel < ItraqConstants::CHANNEL_COUNT[itraq_type_]; ++i_channel)
{
SimTypes::MSSimExperiment::SpectrumType::PeakType p;
// random shift of +-rep_shift around exact position
double rnd_shift = udist(rng_->getTechnicalRng()) * 2 * rep_shift - rep_shift;
p.setMZ(channel_names[itraq_type_].getValue(i_channel, 0) + 0.1 + rnd_shift);
p.setIntensity(itraq_intensity_sum(i_channel, 0));
it->push_back(p);
}
}
}
void ITRAQLabeler::setUpHook(SimTypes::FeatureMapSimVector& features)
{
// no action here .. just check for correct # of channels
Size active_channel_count = 0;
for (ChannelMapType::ConstIterator it = channel_map_.begin(); it != channel_map_.end(); ++it)
{
if (it->second.active)
++active_channel_count;
}
if (features.size() != active_channel_count)
{
throw Exception::IllegalArgument(__FILE__, __LINE__, OPENMS_PRETTY_FUNCTION, String("iTRAQ Labeling received wrong number of channels: ") + String(active_channel_count) + " defined, but " + String(features.size()) + " given as FASTA files.");
}
}
void ICPLLabeler::setUpHook(SimTypes::FeatureMapSimVector& features)
{
// channel check
if (features.size() < 2 || features.size() > 3)
{
throw Exception::IllegalArgument(__FILE__, __LINE__, __PRETTY_FUNCTION__, "We currently support only 2- or 3-channel ICPL");
}
if (param_.getValue("label_proteins") == "true") // loop for protein-labeling (pre-digest-labeling)
{
// first channel labeling (light)
addLabelToProteinHits_(features[0], light_channel_label_);
// second channel labeling (medium)
addLabelToProteinHits_(features[1], medium_channel_label_);
// third channel labeling (heavy) .. if exists
if (features.size() == 3)
{
addLabelToProteinHits_(features[2], heavy_channel_label_);
}
}
}
void SILACLabeler::setUpHook(SimTypes::FeatureMapSimVector& features_to_simulate)
{
// check if we have the correct number of channels
if (features_to_simulate.size() < 2 || features_to_simulate.size() > 3)
{
throw Exception::IllegalArgument(__FILE__, __LINE__, __PRETTY_FUNCTION__, String(features_to_simulate.size()) + " channel(s) given. We currently support only 2-channel SILAC. Please provide two FASTA files!");
}
SimTypes::FeatureMapSim& medium_channel = features_to_simulate[1];
if (medium_channel.getProteinIdentifications().size() > 0)
{
applyLabelToProteinHit_(medium_channel, medium_channel_arginine_label_, medium_channel_lysine_label_);
}
//check for third channel and label
if (features_to_simulate.size() == 3)
{
SimTypes::FeatureMapSim& heavy_channel = features_to_simulate[2];
if (heavy_channel.getProteinIdentifications().size() > 0)
{
applyLabelToProteinHit_(heavy_channel, heavy_channel_arginine_label_, heavy_channel_lysine_label_);
}
}
}
void createTestFeatureMapSimVector_(SimTypes::FeatureMapSimVector& feature_maps, bool add3rd)
{
feature_maps.clear();
SimTypes::FeatureMapSim fm1,fm2,fm3;
ProteinHit prothit1,prothit2,prothit3,prothit4,prothit5,prothit6,prothit7,prothit8,prothit9,prothit10, prothit11, prothit12;
// create first map
prothit1.setSequence("AAAAAAAKAAAAA"); // 2 Fragmente AAAAAAAK und AAAAA und kommt in allen Channels vor
prothit1.setMetaValue("description", "test sequence 1");
prothit1.setAccession("ACC1");
prothit1.setMetaValue("intensity", 200.0);
prothit2.setSequence("CNARCNCNCN"); // 2 Fragmente CNAR und CNCNCN und kommt in allen Channels vor
prothit2.setMetaValue("description", "test sequence 2");
prothit2.setAccession("ACC2");
prothit2.setMetaValue("intensity", 80.0);
prothit3.setSequence("CNHAADDAAAAA"); // ungelabelt, einzelnes Fragment
prothit3.setMetaValue("description", "test sequence 3");
prothit3.setAccession("ACC3");
prothit3.setMetaValue("intensity", 100.0);
prothit12.setSequence("VNAAAAAARVNCNCNAAAA"); // Ergebniss: CNAAAAAAR(Label Medium_R) , CNCNCNAAAA (einmal kommt in allen Channels vor)
prothit12.setMetaValue("description", "test sequence 12");
prothit12.setAccession("ACC5");
prothit12.setMetaValue("intensity", 115.0);
ProteinIdentification protIdent1;
protIdent1.insertHit(prothit1);
protIdent1.insertHit(prothit2);
protIdent1.insertHit(prothit3);
protIdent1.insertHit(prothit12);
vector<ProteinIdentification> protIdents_vec1;
protIdents_vec1.push_back(protIdent1);
fm1.setProteinIdentifications(protIdents_vec1);
// create labeled map
prothit4.setSequence("AAAAAAAKAAAAA"); // Ergbeniss: AAAAAAAK(Label Medium_K) , AAAAA ( einmal kommt in allen Channels vor)
prothit4.setMetaValue("description", "test sequence 4");
prothit4.setAccession("ACC4");
prothit4.setMetaValue("intensity", 50.0);
prothit5.setSequence("CNARCNCNCN"); // Ergebniss: CNAR(Label Medium_R) , CNCNCN (einmal kommt in allen Channels vor)
prothit5.setMetaValue("description", "test sequence 5");
prothit5.setAccession("ACC5");
prothit5.setMetaValue("intensity", 100.0);
prothit6.setSequence("LDRCEL"); // Ergbeniss : LDR(label Medium_R) , CEL (einmal kommt in channel 2 und 3 vor)
prothit6.setMetaValue("description", "test sequence 6");
prothit6.setAccession("ACC6");
prothit6.setMetaValue("intensity", 120.0);
prothit11.setSequence("VNAAAAAARVNCNCNAAAA"); // Ergebniss: CNAAAAAAR(Label Medium_R) , CNCNCNAAAA (einmal kommt in allen Channels vor)
prothit11.setMetaValue("description", "test sequence 11");
prothit11.setAccession("ACC5");
prothit11.setMetaValue("intensity", 110.0);
ProteinIdentification protIdent2;
protIdent2.insertHit(prothit4);
protIdent2.insertHit(prothit5);
protIdent2.insertHit(prothit6);
protIdent2.insertHit(prothit11);
vector<ProteinIdentification> protIdents_vec2;
protIdents_vec2.push_back(protIdent2);
fm2.setProteinIdentifications(protIdents_vec2);
feature_maps.push_back(fm1);
feature_maps.push_back(fm2);
if (add3rd)
{
prothit7.setSequence("AAAAAAAKAAAAA"); // Ergebniss : AAAAAAAK(Label Heavy_K) , AAAAA ( einmal kommt in allen Channels vor )
prothit7.setMetaValue("description", "test sequence 7");
prothit7.setAccession("ACC7");
prothit7.setMetaValue("intensity", 30.0);
prothit8.setSequence("CNARCNCNCN"); // Ergebniss: CNAR(Label Heavy_R) , CNCNCN (einmal kommt in allen Channels vor)
prothit8.setMetaValue("description", "test sequence 8");
prothit8.setAccession("ACC8");
prothit8.setMetaValue("intensity", 130.0);
prothit9.setSequence("LDRCEL"); //Ergebniss: LDR(label Heavy_R) , CEL (einmal kommt in channel 2 und 3 vor)
prothit9.setMetaValue("description", "test sequence 9");
prothit9.setAccession("ACC9");
prothit9.setMetaValue("intensity", 70.0);
prothit10.setSequence("YCYCY"); //Ergebniss: YCYCY kommt nur in diesem Channel vor
prothit10.setMetaValue("description", "test sequence 10");
prothit10.setAccession("ACC10");
prothit10.setMetaValue("intensity", 80.0);
ProteinIdentification protIdent3;
protIdent3.insertHit(prothit7);
protIdent3.insertHit(prothit8);
protIdent3.insertHit(prothit9);
protIdent3.insertHit(prothit10);
vector<ProteinIdentification> protIdents_vec3;
protIdents_vec3.push_back(protIdent3);
fm3.setProteinIdentifications(protIdents_vec3);
feature_maps.push_back(fm3);
}
}
void SILACLabeler::postDigestHook(SimTypes::FeatureMapSimVector& features_to_simulate)
{
SimTypes::FeatureMapSim& light_channel_features = features_to_simulate[0];
SimTypes::FeatureMapSim& medium_channel_features = features_to_simulate[1];
// merge the generated feature maps and create consensus
SimTypes::FeatureMapSim final_feature_map = mergeProteinIdentificationsMaps_(features_to_simulate);
if (features_to_simulate.size() == 2)
{
Map<String, Feature> unlabeled_features_index;
for (SimTypes::FeatureMapSim::iterator unlabeled_features_iter = light_channel_features.begin();
unlabeled_features_iter != light_channel_features.end();
++unlabeled_features_iter)
{
(*unlabeled_features_iter).ensureUniqueId();
unlabeled_features_index.insert(std::make_pair(
(*unlabeled_features_iter).getPeptideIdentifications()[0].getHits()[0].getSequence().toString()
,
*unlabeled_features_iter
));
}
// iterate over second map
for (SimTypes::FeatureMapSim::iterator labeled_feature_iter = medium_channel_features.begin(); labeled_feature_iter != medium_channel_features.end(); ++labeled_feature_iter)
{
const String unmodified_sequence = getUnmodifiedSequence_(*labeled_feature_iter, medium_channel_arginine_label_, medium_channel_lysine_label_);
// guarantee uniqueness
(*labeled_feature_iter).ensureUniqueId();
// check if we have a pair
if (unlabeled_features_index.has(unmodified_sequence))
{
// own scope as we don't know what happens to 'f_modified' once we call erase() below
Feature& unlabeled_feature = unlabeled_features_index[unmodified_sequence];
// guarantee uniqueness
unlabeled_feature.ensureUniqueId();
// feature has a SILAC Label and is not equal to non-labeled
if ((*labeled_feature_iter).getPeptideIdentifications()[0].getHits()[0].getSequence().isModified())
{
// add features to final map
final_feature_map.push_back(*labeled_feature_iter);
final_feature_map.push_back(unlabeled_feature);
// create consensus feature
ConsensusFeature cf;
cf.insert(MEDIUM_FEATURE_MAPID_, *labeled_feature_iter);
cf.insert(LIGHT_FEATURE_MAPID_, unlabeled_feature);
cf.ensureUniqueId();
consensus_.push_back(cf);
// remove unlabeled feature
unlabeled_features_index.erase(unmodified_sequence);
}
else
{
// merge features since they are equal
Feature final_feature = mergeFeatures_(*labeled_feature_iter, unmodified_sequence, unlabeled_features_index, 1, 2);
final_feature_map.push_back(final_feature);
}
}
else // no SILAC pair, just add the labeled one
{
final_feature_map.push_back(*labeled_feature_iter);
}
}
// add singletons from unlabeled channel
// clean up unlabeled_index
for (Map<String, Feature>::iterator unlabeled_index_iter = unlabeled_features_index.begin(); unlabeled_index_iter != unlabeled_features_index.end(); ++unlabeled_index_iter)
{
// the single ones from c0
final_feature_map.push_back(unlabeled_index_iter->second);
}
}
// merge three channels
if (features_to_simulate.size() == 3)
{
// index of unlabeled channelunlabeled_feature
Map<String, Feature> unlabeled_features_index;
for (SimTypes::FeatureMapSim::iterator unlabeled_features_iter = light_channel_features.begin();
unlabeled_features_iter != light_channel_features.end();
++unlabeled_features_iter)
{
(*unlabeled_features_iter).ensureUniqueId();
unlabeled_features_index.insert(std::make_pair(
(*unlabeled_features_iter).getPeptideIdentifications()[0].getHits()[0].getSequence().toString()
,
*unlabeled_features_iter
));
}
// index of labeled channel
Map<String, Feature> medium_features_index;
for (SimTypes::FeatureMapSim::iterator labeled_features_iter = medium_channel_features.begin();
labeled_features_iter != medium_channel_features.end();
++labeled_features_iter)
{
(*labeled_features_iter).ensureUniqueId();
medium_features_index.insert(std::make_pair(
getUnmodifiedSequence_(*labeled_features_iter, medium_channel_arginine_label_, medium_channel_lysine_label_)
,
*labeled_features_iter
));
}
SimTypes::FeatureMapSim& heavy_labeled_features = features_to_simulate[2];
for (SimTypes::FeatureMapSim::iterator heavy_labeled_feature_iter = heavy_labeled_features.begin();
heavy_labeled_feature_iter != heavy_labeled_features.end();
++heavy_labeled_feature_iter)
{
Feature& heavy_feature = *heavy_labeled_feature_iter;
heavy_feature.ensureUniqueId();
String heavy_feature_unmodified_sequence = getUnmodifiedSequence_(heavy_feature, heavy_channel_arginine_label_, heavy_channel_lysine_label_);
if (unlabeled_features_index.has(heavy_feature_unmodified_sequence) && medium_features_index.has(heavy_feature_unmodified_sequence))
{
// it is a triplet
// c2 & c1 modified
if (heavy_feature_unmodified_sequence != heavy_feature.getPeptideIdentifications()[0].getHits()[0].getSequence().toString())
{
// add features to final map
final_feature_map.push_back(heavy_feature);
final_feature_map.push_back(medium_features_index[heavy_feature_unmodified_sequence]);
final_feature_map.push_back(unlabeled_features_index[heavy_feature_unmodified_sequence]);
ConsensusFeature c_triplet;
c_triplet.insert(HEAVY_FEATURE_MAPID_, heavy_feature);
c_triplet.insert(LIGHT_FEATURE_MAPID_, unlabeled_features_index[heavy_feature_unmodified_sequence]);
c_triplet.insert(MEDIUM_FEATURE_MAPID_, medium_features_index[heavy_feature_unmodified_sequence]);
c_triplet.ensureUniqueId();
consensus_.push_back(c_triplet);
}
else
{
// merge all three channels
Feature completeMerge = mergeAllChannelFeatures_(heavy_feature, heavy_feature_unmodified_sequence, unlabeled_features_index, medium_features_index);
final_feature_map.push_back(completeMerge);
}
// remove features from indices
unlabeled_features_index.erase(heavy_feature_unmodified_sequence);
medium_features_index.erase(heavy_feature_unmodified_sequence);
}
else if (unlabeled_features_index.has(heavy_feature_unmodified_sequence))
{
// 2nd case light and heavy pair
if (heavy_feature_unmodified_sequence != heavy_feature.getPeptideIdentifications()[0].getHits()[0].getSequence().toString())
{
// add features to final map
final_feature_map.push_back(heavy_feature);
final_feature_map.push_back(unlabeled_features_index[heavy_feature_unmodified_sequence]);
ConsensusFeature c_duplet;
c_duplet.insert(HEAVY_FEATURE_MAPID_, heavy_feature);
c_duplet.insert(LIGHT_FEATURE_MAPID_, unlabeled_features_index[heavy_feature_unmodified_sequence]);
c_duplet.ensureUniqueId();
consensus_.push_back(c_duplet);
}
else
{
// merge all three channels
Feature completeMerge = mergeFeatures_(heavy_feature, heavy_feature_unmodified_sequence, unlabeled_features_index, 1, 3);
final_feature_map.push_back(completeMerge);
}
// remove features from indices
unlabeled_features_index.erase(heavy_feature_unmodified_sequence);
}
else if (medium_features_index.has(heavy_feature_unmodified_sequence))
{
// 3rd case medium and heavy pair
if (heavy_feature_unmodified_sequence != heavy_feature.getPeptideIdentifications()[0].getHits()[0].getSequence().toString())
{
// add features to final map
final_feature_map.push_back(heavy_feature);
final_feature_map.push_back(medium_features_index[heavy_feature_unmodified_sequence]);
ConsensusFeature c_duplet;
c_duplet.insert(HEAVY_FEATURE_MAPID_, heavy_feature);
c_duplet.insert(MEDIUM_FEATURE_MAPID_, medium_features_index[heavy_feature_unmodified_sequence]);
c_duplet.ensureUniqueId();
consensus_.push_back(c_duplet);
}
else
{
// merge all
Feature completeMerge = mergeFeatures_(heavy_feature, heavy_feature_unmodified_sequence, medium_features_index, 2, 3);
final_feature_map.push_back(completeMerge);
}
// remove features from indices
medium_features_index.erase(heavy_feature_unmodified_sequence);
}
else
{
// heavy feature is a singleton
final_feature_map.push_back(heavy_feature);
}
}
// clean up labeled_index
for (Map<String, Feature>::iterator medium_channle_index_iterator = medium_features_index.begin(); medium_channle_index_iterator != medium_features_index.end(); ++medium_channle_index_iterator)
{
Feature& medium_channel_feature = medium_channle_index_iterator->second;
medium_channel_feature.ensureUniqueId();
String medium_channel_feature_unmodified_sequence = getUnmodifiedSequence_(medium_channel_feature, medium_channel_arginine_label_, medium_channel_lysine_label_);
if (unlabeled_features_index.has(medium_channel_feature_unmodified_sequence))
{
// 1. case: pair between c0 and c1
if (medium_channel_feature.getPeptideIdentifications()[0].getHits()[0].getSequence().isModified())
{
// add features to final map
final_feature_map.push_back(medium_channel_feature);
final_feature_map.push_back(unlabeled_features_index[medium_channel_feature_unmodified_sequence]);
ConsensusFeature c_duplet;
c_duplet.insert(MEDIUM_FEATURE_MAPID_, medium_channel_feature);
c_duplet.insert(LIGHT_FEATURE_MAPID_, unlabeled_features_index[medium_channel_feature_unmodified_sequence]);
c_duplet.ensureUniqueId();
consensus_.push_back(c_duplet);
}
else
{
// merge
Feature completeMerge = mergeFeatures_(medium_channel_feature, medium_channel_feature_unmodified_sequence, unlabeled_features_index, 1, 2);
final_feature_map.push_back(completeMerge);
}
// remove features from indices
unlabeled_features_index.erase(medium_channel_feature_unmodified_sequence);
}
else
{
// c1 is alone
final_feature_map.push_back(medium_channel_feature);
}
}
// clean up unlabeled_index
for (Map<String, Feature>::iterator unlabeled_index_iter = unlabeled_features_index.begin(); unlabeled_index_iter != unlabeled_features_index.end(); ++unlabeled_index_iter)
{
// the single ones from c0
final_feature_map.push_back(unlabeled_index_iter->second);
}
}
features_to_simulate.clear();
features_to_simulate.push_back(final_feature_map);
consensus_.setProteinIdentifications(final_feature_map.getProteinIdentifications());
ConsensusMap::FileDescription map_description;
map_description.label = "Simulation (Labeling Consensus)";
map_description.size = features_to_simulate.size();
consensus_.getFileDescriptions()[0] = map_description;
}
void ICPLLabeler::postDigestHook(SimTypes::FeatureMapSimVector& features_to_simulate)
{
SimTypes::FeatureMapSim& light_labeled_features = features_to_simulate[0];
SimTypes::FeatureMapSim& medium_labeled_features = features_to_simulate[1];
if (param_.getValue("label_proteins") == "false") // loop for peptide-labeling (post-digest-labeling)
{
// iterate over first map for light labeling
for (SimTypes::FeatureMapSim::iterator lf_iter = light_labeled_features.begin(); lf_iter != light_labeled_features.end(); ++lf_iter)
{
lf_iter->ensureUniqueId();
addModificationToPeptideHit_(*lf_iter, light_channel_label_);
}
// iterate over second map for medium labeling
for (SimTypes::FeatureMapSim::iterator lf_iter = medium_labeled_features.begin(); lf_iter != medium_labeled_features.end(); ++lf_iter)
{
lf_iter->ensureUniqueId();
addModificationToPeptideHit_(*lf_iter, medium_channel_label_);
}
if (features_to_simulate.size() == 3) //third channel labeling can only be done, if a third channel exist
{
SimTypes::FeatureMapSim& heavy_labeled_features = features_to_simulate[2];
// iterate over third map
for (SimTypes::FeatureMapSim::iterator lf_iter = heavy_labeled_features.begin(); lf_iter != heavy_labeled_features.end(); ++lf_iter)
{
lf_iter->ensureUniqueId();
addModificationToPeptideHit_(*lf_iter, heavy_channel_label_);
}
}
}
// merge the generated feature maps and create consensus
SimTypes::FeatureMapSim final_feature_map = mergeProteinIdentificationsMaps_(features_to_simulate);
if (features_to_simulate.size() == 2) // merge_modus for two FeatureMaps
{
// create index of light channel features for easy mapping of medium-to-light channel
Map<String, Feature> light_labeled_features_index;
for (SimTypes::FeatureMapSim::iterator light_labeled_features_iter = light_labeled_features.begin();
light_labeled_features_iter != light_labeled_features.end();
++light_labeled_features_iter)
{
(*light_labeled_features_iter).ensureUniqueId();
light_labeled_features_index.insert(std::make_pair(
getUnmodifiedAASequence_((*light_labeled_features_iter), light_channel_label_),
*light_labeled_features_iter
));
}
// iterate over second map
for (SimTypes::FeatureMapSim::iterator medium_labeled_feature_iter = medium_labeled_features.begin(); medium_labeled_feature_iter != medium_labeled_features.end(); ++medium_labeled_feature_iter)
{
AASequence medium_labeled_feature_sequence = (*medium_labeled_feature_iter).getPeptideIdentifications()[0].getHits()[0].getSequence();
// guarantee uniqueness
(*medium_labeled_feature_iter).ensureUniqueId();
// check if we have a pair
if (light_labeled_features_index.has(getUnmodifiedAASequence_((*medium_labeled_feature_iter), medium_channel_label_)))
{
// own scope as we don't know what happens to 'f_modified' once we call erase() below
Feature& light_labeled_feature = light_labeled_features_index[getUnmodifiedAASequence_((*medium_labeled_feature_iter), medium_channel_label_)];
// guarantee uniqueness
light_labeled_feature.ensureUniqueId();
if (medium_labeled_feature_sequence.isModified()) // feature has a medium ICPL-Label and is not equal to light-labeled
{
// add features to final map
final_feature_map.push_back(*medium_labeled_feature_iter);
final_feature_map.push_back(light_labeled_feature);
// create consensus feature
ConsensusFeature cf;
cf.insert(MEDIUM_FEATURE_MAPID_, *medium_labeled_feature_iter);
cf.insert(LIGHT_FEATURE_MAPID_, light_labeled_feature);
consensus_.push_back(cf);
// remove light-labeled feature
light_labeled_features_index.erase(getUnmodifiedAASequence_((*medium_labeled_feature_iter), medium_channel_label_));
}
else
{
// merge features since they are equal
Feature final_feature = mergeFeatures_(*medium_labeled_feature_iter, medium_labeled_feature_sequence, light_labeled_features_index);
final_feature_map.push_back(final_feature);
}
}
else // no ICPL pair, just add the medium-labeled one
{
final_feature_map.push_back(*medium_labeled_feature_iter);
}
}
// add singletons from light-labeled channel
// clean up light-labeled_index
for (Map<String, Feature>::iterator light_labeled_index_iter = light_labeled_features_index.begin(); light_labeled_index_iter != light_labeled_features_index.end(); ++light_labeled_index_iter)
{
// the single ones from c0
final_feature_map.push_back(light_labeled_index_iter->second);
}
}
else if (features_to_simulate.size() == 3) // merge_modus for three Channels
{
// create index of light channel features for easy mapping of heavy-to-medium-to-light channel
Map<String, Feature> light_labeled_features_index;
for (SimTypes::FeatureMapSim::iterator light_labeled_features_iter = light_labeled_features.begin();
light_labeled_features_iter != light_labeled_features.end();
++light_labeled_features_iter)
{
(*light_labeled_features_iter).ensureUniqueId();
light_labeled_features_index.insert(std::make_pair(
getUnmodifiedAASequence_(*light_labeled_features_iter, light_channel_label_),
*light_labeled_features_iter
));
}
// create index of medium channel features for easy mapping of heavy-to-medium-to-light channel
Map<String, Feature> medium_labeled_features_index;
for (SimTypes::FeatureMapSim::iterator medium_labeled_features_iter = medium_labeled_features.begin();
medium_labeled_features_iter != medium_labeled_features.end();
++medium_labeled_features_iter)
{
(*medium_labeled_features_iter).ensureUniqueId();
medium_labeled_features_index.insert(std::make_pair(
getUnmodifiedAASequence_((*medium_labeled_features_iter), medium_channel_label_),
*medium_labeled_features_iter
));
}
for (SimTypes::FeatureMapSim::iterator heavy_labeled_feature_iter = features_to_simulate[2].begin(); heavy_labeled_feature_iter != features_to_simulate[2].end(); ++heavy_labeled_feature_iter)
{
Feature& heavy_feature = *heavy_labeled_feature_iter;
String heavy_feature_unmodified_sequence = getUnmodifiedAASequence_(heavy_feature, heavy_channel_label_);
heavy_feature.ensureUniqueId();
if (light_labeled_features_index.has(heavy_feature_unmodified_sequence) && medium_labeled_features_index.has(heavy_feature_unmodified_sequence))
{
// 1st case .. it is a triplet
if (heavy_feature.getPeptideIdentifications()[0].getHits()[0].getSequence().isModified())
{
// if heavy feature is modified, all light and medium channel are also
// add features to final map
final_feature_map.push_back(heavy_feature);
final_feature_map.push_back(medium_labeled_features_index[heavy_feature_unmodified_sequence]);
final_feature_map.push_back(light_labeled_features_index[heavy_feature_unmodified_sequence]);
// create triplet consensus feature
ConsensusFeature c_triplet;
c_triplet.insert(HEAVY_FEATURE_MAPID_, heavy_feature);
c_triplet.insert(LIGHT_FEATURE_MAPID_, light_labeled_features_index[heavy_feature_unmodified_sequence]);
c_triplet.insert(MEDIUM_FEATURE_MAPID_, medium_labeled_features_index[heavy_feature_unmodified_sequence]);
consensus_.push_back(c_triplet);
}
else
{
// merge all three channels
Feature c2c1 = mergeFeatures_(heavy_feature, AASequence::fromString(heavy_feature_unmodified_sequence), medium_labeled_features_index);
Feature completeMerge = mergeFeatures_(c2c1, AASequence::fromString(heavy_feature_unmodified_sequence), light_labeled_features_index);
final_feature_map.push_back(completeMerge);
}
// remove features from indices
light_labeled_features_index.erase(heavy_feature_unmodified_sequence);
medium_labeled_features_index.erase(heavy_feature_unmodified_sequence);
}
else if (light_labeled_features_index.has(heavy_feature_unmodified_sequence))
{
// 2.Fall -> c0 - c2
if (heavy_feature.getPeptideIdentifications()[0].getHits()[0].getSequence().isModified())
{
// add features to final map
final_feature_map.push_back(heavy_feature);
final_feature_map.push_back(light_labeled_features_index[heavy_feature_unmodified_sequence]);
ConsensusFeature c_triplet;
c_triplet.insert(HEAVY_FEATURE_MAPID_, heavy_feature);
c_triplet.insert(LIGHT_FEATURE_MAPID_, light_labeled_features_index[heavy_feature_unmodified_sequence]);
consensus_.push_back(c_triplet);
}
else
{
// merge all three channels
Feature completeMerge = mergeFeatures_(heavy_feature, AASequence::fromString(heavy_feature_unmodified_sequence), light_labeled_features_index);
final_feature_map.push_back(completeMerge);
}
// remove features from indices
light_labeled_features_index.erase(heavy_feature_unmodified_sequence);
}
else if (medium_labeled_features_index.has(heavy_feature_unmodified_sequence))
{
// 3.Fall -> c1 - c2
if (heavy_feature.getPeptideIdentifications()[0].getHits()[0].getSequence().isModified())
{
// add features to final map
final_feature_map.push_back(heavy_feature);
final_feature_map.push_back(medium_labeled_features_index[heavy_feature_unmodified_sequence]);
ConsensusFeature c_triplet;
c_triplet.insert(HEAVY_FEATURE_MAPID_, heavy_feature);
c_triplet.insert(MEDIUM_FEATURE_MAPID_, medium_labeled_features_index[heavy_feature_unmodified_sequence]);
consensus_.push_back(c_triplet);
}
else
{
// merge all
Feature completeMerge = mergeFeatures_(heavy_feature, AASequence::fromString(heavy_feature_unmodified_sequence), medium_labeled_features_index);
final_feature_map.push_back(completeMerge);
}
// remove features from indices
medium_labeled_features_index.erase(heavy_feature_unmodified_sequence);
}
else
{
// 4.Fall -> alleine
final_feature_map.push_back(heavy_feature);
}
}
// clean up medium-labeled_index
for (Map<String, Feature>::iterator medium_labeled_index_iter = medium_labeled_features_index.begin(); medium_labeled_index_iter != medium_labeled_features_index.end(); ++medium_labeled_index_iter)
{
Feature& medium_labeled_feature = medium_labeled_index_iter->second;
medium_labeled_feature.ensureUniqueId();
String medium_labeled_feature_unmodified_sequence = medium_labeled_feature.getPeptideIdentifications()[0].getHits()[0].getSequence().toUnmodifiedString();
if (light_labeled_features_index.has(medium_labeled_feature_unmodified_sequence))
{
// 1. case: pair between c0 and c1
if (medium_labeled_feature.getPeptideIdentifications()[0].getHits()[0].getSequence().isModified())
{
// add features to final map
final_feature_map.push_back(medium_labeled_feature);
final_feature_map.push_back(light_labeled_features_index[medium_labeled_feature_unmodified_sequence]);
ConsensusFeature c_triplet;
c_triplet.insert(MEDIUM_FEATURE_MAPID_, medium_labeled_feature);
c_triplet.insert(LIGHT_FEATURE_MAPID_, light_labeled_features_index[medium_labeled_feature_unmodified_sequence]);
consensus_.push_back(c_triplet);
}
else
{
// merge
Feature completeMerge = mergeFeatures_(medium_labeled_feature, AASequence::fromString(medium_labeled_feature_unmodified_sequence), light_labeled_features_index);
final_feature_map.push_back(completeMerge);
}
// remove features from indices
light_labeled_features_index.erase(medium_labeled_feature_unmodified_sequence);
}
else
{
// c1 is alone
final_feature_map.push_back(medium_labeled_feature);
}
}
// clean up light-labeled_index
for (Map<String, Feature>::iterator light_labeled_index_iter = light_labeled_features_index.begin(); light_labeled_index_iter != light_labeled_features_index.end(); ++light_labeled_index_iter)
{
// the single ones from c0
final_feature_map.push_back(light_labeled_index_iter->second);
}
}
features_to_simulate.clear();
features_to_simulate.push_back(final_feature_map);
consensus_.setProteinIdentifications(final_feature_map.getProteinIdentifications());
ConsensusMap::FileDescription map_description;
map_description.label = "Simulation (Labeling Consensus)";
map_description.size = features_to_simulate.size();
consensus_.getFileDescriptions()[0] = map_description;
}