void BaseLabeler::recomputeConsensus_(const FeatureMapSim & simulated_features)
{
// iterate over all given features stored in the labeling consensus and try to find the corresponding feature in
// in the feature map
// build index for faster access
Map<String, IntList> id_map;
Map<UInt64, Size> features_per_labeled_map;
for (Size i = 0; i < simulated_features.size(); ++i)
{
if (simulated_features[i].metaValueExists("parent_feature"))
{
LOG_DEBUG << "Checking [" << i << "]: " << simulated_features[i].getPeptideIdentifications()[0].getHits()[0].getSequence().toString()
<< " with charge " << simulated_features[i].getCharge() << " (" << simulated_features[i].getMetaValue("charge_adducts") << ")"
<< " parent was " << simulated_features[i].getMetaValue("parent_feature") << std::endl;
id_map[simulated_features[i].getMetaValue("parent_feature")].push_back((Int)i);
UInt64 map_index = 0;
if (simulated_features[i].metaValueExists("map_index"))
{
map_index = simulated_features[i].getMetaValue("map_index");
}
++features_per_labeled_map[map_index];
}
}
for (Map<String, IntList>::iterator it = id_map.begin(); it != id_map.end(); ++it)
{
LOG_DEBUG << it->first << " " << it->second << std::endl;
}
// new consensus map
ConsensusMap new_cm;
// initialize submaps in consensus map
for (Map<UInt64, Size>::Iterator it = features_per_labeled_map.begin(); it != features_per_labeled_map.end(); ++it)
{
new_cm.getFileDescriptions()[it->first].size = it->second;
new_cm.getFileDescriptions()[it->first].unique_id = simulated_features.getUniqueId();
}
for (ConsensusMap::iterator cm_iter = consensus_.begin(); cm_iter != consensus_.end(); ++cm_iter)
{
bool complete = true;
LOG_DEBUG << "Checking consensus feature containing: " << std::endl;
// check if we have all elements of current CF in the new feature map (simulated_features)
for (ConsensusFeature::iterator cf_iter = (*cm_iter).begin(); cf_iter != (*cm_iter).end(); ++cf_iter)
{
complete &= id_map.has(String((*cf_iter).getUniqueId()));
LOG_DEBUG << "\t" << String((*cf_iter).getUniqueId()) << std::endl;
}
if (complete)
{
// get all elements sorted by charge state; since the same charge can be achieved by different
// adduct compositions we use the adduct-string as indicator to find the groups
Map<String, std::set<FeatureHandle, FeatureHandle::IndexLess> > charge_mapping;
for (ConsensusFeature::iterator cf_iter = (*cm_iter).begin(); cf_iter != (*cm_iter).end(); ++cf_iter)
{
IntList feature_indices = id_map[String((*cf_iter).getUniqueId())];
for (IntList::iterator it = feature_indices.begin(); it != feature_indices.end(); ++it)
{
UInt64 map_index = 0;
if (simulated_features[*it].metaValueExists("map_index"))
{
map_index = simulated_features[*it].getMetaValue("map_index");
}
if (charge_mapping.has(simulated_features[*it].getMetaValue("charge_adducts")))
{
charge_mapping[simulated_features[*it].getMetaValue("charge_adducts")].insert(FeatureHandle(map_index, simulated_features[*it]));
}
else
{
LOG_DEBUG << "Create new set with charge composition " << simulated_features[*it].getMetaValue("charge_adducts") << std::endl;
std::set<FeatureHandle, FeatureHandle::IndexLess> fh_set;
fh_set.insert(FeatureHandle(map_index, simulated_features[*it]));
charge_mapping.insert(std::make_pair(simulated_features[*it].getMetaValue("charge_adducts"), fh_set));
}
}
}
// create new consensus feature from derived features (separated by charge, if charge != 0)
for (Map<String, std::set<FeatureHandle, FeatureHandle::IndexLess> >::const_iterator charge_group_it = charge_mapping.begin();
charge_group_it != charge_mapping.end();
++charge_group_it)
{
ConsensusFeature cf;
cf.setCharge((*(*charge_group_it).second.begin()).getCharge());
cf.setMetaValue("charge_adducts", charge_group_it->first);
std::vector<PeptideIdentification> ids;
for (std::set<FeatureHandle, FeatureHandle::IndexLess>::const_iterator fh_it = (charge_group_it->second).begin(); fh_it != (charge_group_it->second).end(); ++fh_it)
{
cf.insert(*fh_it);
// append identifications
Size f_index = simulated_features.uniqueIdToIndex(fh_it->getUniqueId());
std::vector<PeptideIdentification> ids_feature = simulated_features[f_index].getPeptideIdentifications();
ids.insert(ids.end(), ids_feature.begin(), ids_feature.end());
}
cf.computeMonoisotopicConsensus();
cf.setPeptideIdentifications(ids);
new_cm.push_back(cf);
}
}
}
new_cm.setProteinIdentifications(simulated_features.getProteinIdentifications());
consensus_.swap(new_cm);
consensus_.applyMemberFunction(&UniqueIdInterface::ensureUniqueId);
}
ExitCodes main_(int, const char**) override
{
//-------------------------------------------------------------
// parameter handling
//-------------------------------------------------------------
String in = getStringOption_("in");
String out = getStringOption_("out");
FileTypes::Type out_type = FileTypes::nameToType(getStringOption_("out_type"));
if (out_type == FileTypes::UNKNOWN)
{
out_type = FileHandler().getTypeByFileName(out);
}
//-------------------------------------------------------------
// loading input
//-------------------------------------------------------------
MzMLFile mz_data_file;
mz_data_file.setLogType(log_type_);
PeakMap ms_peakmap;
std::vector<Int> ms_level(1, 1);
(mz_data_file.getOptions()).setMSLevels(ms_level);
mz_data_file.load(in, ms_peakmap);
if (ms_peakmap.size() == 0)
{
LOG_WARN << "The given file does not contain any conventional peak data, but might"
" contain chromatograms. This tool currently cannot handle them, sorry.";
return INCOMPATIBLE_INPUT_DATA;
}
// make sure that the spectra are sorted by m/z
ms_peakmap.sortSpectra(true);
//-------------------------------------------------------------
// get params for MTD and EPD algorithms
//-------------------------------------------------------------
Param com_param = getParam_().copy("algorithm:common:", true);
writeDebug_("Common parameters passed to both sub-algorithms (mtd and epd)", com_param, 3);
Param mtd_param = getParam_().copy("algorithm:mtd:", true);
writeDebug_("Parameters passed to MassTraceDetection", mtd_param, 3);
Param epd_param = getParam_().copy("algorithm:epd:", true);
writeDebug_("Parameters passed to ElutionPeakDetection", epd_param, 3);
//-------------------------------------------------------------
// configure and run MTD
//-------------------------------------------------------------
MassTraceDetection mt_ext;
mtd_param.insert("", com_param);
mtd_param.remove("chrom_fwhm");
mt_ext.setParameters(mtd_param);
vector<MassTrace> m_traces;
mt_ext.run(ms_peakmap, m_traces);
vector<MassTrace> m_traces_final;
bool use_epd = epd_param.getValue("enabled").toBool();
if (!use_epd)
{
swap(m_traces_final, m_traces);
}
else
{
ElutionPeakDetection ep_det;
epd_param.remove("enabled"); // artificially added above
epd_param.insert("", com_param);
ep_det.setParameters(epd_param);
std::vector<MassTrace> split_mtraces;
// note: this step will destroy any meta data annotation (e.g. FWHM_mz_avg)
ep_det.detectPeaks(m_traces, split_mtraces);
if (ep_det.getParameters().getValue("width_filtering") == "auto")
{
m_traces_final.clear();
ep_det.filterByPeakWidth(split_mtraces, m_traces_final);
LOG_INFO << "Notice: " << split_mtraces.size() - m_traces_final.size()
<< " of total " << split_mtraces.size()
<< " were dropped because of too low peak width." << std::endl;
}
else
{
swap(m_traces_final, split_mtraces);
}
}
//-------------------------------------------------------------
// writing consensus map output
//-------------------------------------------------------------
if (out_type == FileTypes::CONSENSUSXML)
{
ConsensusMap consensus_map;
StringList ms_runs;
ms_peakmap.getPrimaryMSRunPath(ms_runs);
consensus_map.setPrimaryMSRunPath(ms_runs);
for (Size i = 0; i < m_traces_final.size(); ++i)
{
if (m_traces_final[i].getSize() == 0) continue;
ConsensusFeature fcons;
int k = 0;
for (MassTrace::const_iterator it = m_traces_final[i].begin(); it != m_traces_final[i].end(); ++it)
{
FeatureHandle fhandle;
fhandle.setRT(it->getRT());
fhandle.setMZ(it->getMZ());
fhandle.setIntensity(it->getIntensity());
fhandle.setUniqueId(++k);
fcons.insert(fhandle);
}
fcons.setMetaValue(3, m_traces_final[i].getLabel());
fcons.setCharge(0);
fcons.setWidth(m_traces_final[i].estimateFWHM(use_epd));
fcons.setQuality(1 - (1.0 / m_traces_final[i].getSize()));
fcons.setRT(m_traces_final[i].getCentroidRT());
fcons.setMZ(m_traces_final[i].getCentroidMZ());
fcons.setIntensity(m_traces_final[i].getIntensity(false));
consensus_map.push_back(fcons);
}
consensus_map.applyMemberFunction(&UniqueIdInterface::setUniqueId);
addDataProcessing_(consensus_map, getProcessingInfo_(DataProcessing::QUANTITATION));
consensus_map.setUniqueId();
ConsensusXMLFile().store(out, consensus_map);
}
else //(out_type == FileTypes::FEATUREXML)
{
//-----------------------------------------------------------
// convert mass traces to features
//-----------------------------------------------------------
std::vector<double> stats_sd;
FeatureMap ms_feat_map;
StringList ms_runs;
ms_peakmap.getPrimaryMSRunPath(ms_runs);
ms_feat_map.setPrimaryMSRunPath(ms_runs);
for (Size i = 0; i < m_traces_final.size(); ++i)
{
if (m_traces_final[i].getSize() == 0) continue;
m_traces_final[i].updateMeanMZ();
m_traces_final[i].updateWeightedMZsd();
Feature f;
f.setMetaValue(3, m_traces_final[i].getLabel());
f.setCharge(0);
f.setMZ(m_traces_final[i].getCentroidMZ());
f.setIntensity(m_traces_final[i].getIntensity(false));
f.setRT(m_traces_final[i].getCentroidRT());
f.setWidth(m_traces_final[i].estimateFWHM(use_epd));
f.setOverallQuality(1 - (1.0 / m_traces_final[i].getSize()));
f.getConvexHulls().push_back(m_traces_final[i].getConvexhull());
double sd = m_traces_final[i].getCentroidSD();
f.setMetaValue("SD", sd);
f.setMetaValue("SD_ppm", sd / f.getMZ() * 1e6);
if (m_traces_final[i].fwhm_mz_avg > 0) f.setMetaValue("FWHM_mz_avg", m_traces_final[i].fwhm_mz_avg);
stats_sd.push_back(m_traces_final[i].getCentroidSD());
ms_feat_map.push_back(f);
}
// print some stats about standard deviation of mass traces
if (stats_sd.size() > 0)
{
std::sort(stats_sd.begin(), stats_sd.end());
LOG_INFO << "Mass trace m/z s.d.\n"
<< " low quartile: " << stats_sd[stats_sd.size() * 1 / 4] << "\n"
<< " median: " << stats_sd[stats_sd.size() * 1 / 2] << "\n"
<< " upp quartile: " << stats_sd[stats_sd.size() * 3 / 4] << std::endl;
}
ms_feat_map.applyMemberFunction(&UniqueIdInterface::setUniqueId);
//-------------------------------------------------------------
// writing output
//-------------------------------------------------------------
// annotate output with data processing info TODO
addDataProcessing_(ms_feat_map, getProcessingInfo_(DataProcessing::QUANTITATION));
//ms_feat_map.setUniqueId();
FeatureXMLFile().store(out, ms_feat_map);
}
return EXECUTION_OK;
}
void EDTAFile::load(const String& filename, ConsensusMap& consensus_map)
{
// load input
TextFile input(filename);
TextFile::ConstIterator input_it = input.begin();
// reset map
consensus_map = ConsensusMap();
consensus_map.setUniqueId();
char separator = ' ';
if (input_it->hasSubstring("\t"))
separator = '\t';
else if (input_it->hasSubstring(" "))
separator = ' ';
else if (input_it->hasSubstring(","))
separator = ',';
// parsing header line
std::vector<String> headers;
input_it->split(separator, headers);
int offset = 0;
for (Size i = 0; i < headers.size(); ++i)
{
headers[i].trim();
}
String header_trimmed = *input.begin();
header_trimmed.trim();
enum
{
TYPE_UNDEFINED,
TYPE_OLD_NOCHARGE,
TYPE_OLD_CHARGE,
TYPE_CONSENSUS
}
input_type = TYPE_UNDEFINED;
Size input_features = 1;
double rt = 0.0;
double mz = 0.0;
double it = 0.0;
Int ch = 0;
if (headers.size() <= 2)
{
throw Exception::ParseError(__FILE__, __LINE__, __PRETTY_FUNCTION__, "", String("Failed parsing in line 1: not enough columns! Expected at least 3 columns!\nOffending line: '") + header_trimmed + "' (line 1)\n");
}
else if (headers.size() == 3)
input_type = TYPE_OLD_NOCHARGE;
else if (headers.size() == 4)
input_type = TYPE_OLD_CHARGE;
// see if we have a header
try
{
// try to convert... if not: thats a header
rt = headers[0].toDouble();
mz = headers[1].toDouble();
it = headers[2].toDouble();
}
catch (Exception::BaseException&)
{
offset = 1;
++input_it;
LOG_INFO << "Detected a header line.\n";
}
if (headers.size() >= 5)
{
if (String(headers[4].trim()).toUpper() == "RT1")
input_type = TYPE_CONSENSUS;
else
input_type = TYPE_OLD_CHARGE;
}
if (input_type == TYPE_CONSENSUS)
{
// Every consensus style line includes features with four columns.
// The remainder is meta data
input_features = headers.size() / 4;
}
if (offset == 0 && (input_type == TYPE_OLD_CHARGE || input_type == TYPE_CONSENSUS))
{
throw Exception::ParseError(__FILE__, __LINE__, __PRETTY_FUNCTION__, "", String("Failed parsing in line 1: No HEADER provided. This is only allowed for three columns. You have more!\nOffending line: '") + header_trimmed + "' (line 1)\n");
}
SignedSize input_size = input.end() - input.begin();
ConsensusMap::FileDescription desc;
desc.filename = filename;
desc.size = (input_size) - offset;
consensus_map.getFileDescriptions()[0] = desc;
// parsing features
consensus_map.reserve(input_size);
for (; input_it != input.end(); ++input_it)
{
//do nothing for empty lines
String line_trimmed = *input_it;
line_trimmed.trim();
if (line_trimmed == "")
{
if ((input_it - input.begin()) < input_size - 1) LOG_WARN << "Notice: Empty line ignored (line " << ((input_it - input.begin()) + 1) << ").";
continue;
}
//split line to tokens
std::vector<String> parts;
input_it->split(separator, parts);
//abort if line does not contain enough fields
if (parts.size() < 3)
{
throw Exception::ParseError(__FILE__, __LINE__, __PRETTY_FUNCTION__, "",
String("Failed parsing in line ")
+ String((input_it - input.begin()) + 1)
+ ": At least three columns are needed! (got "
+ String(parts.size())
+ ")\nOffending line: '"
+ line_trimmed
+ "' (line "
+ String((input_it - input.begin()) + 1)
+ ")\n");
}
ConsensusFeature cf;
cf.setUniqueId();
try
{
// Convert values. Will return -1 if not available.
rt = checkedToDouble_(parts, 0);
mz = checkedToDouble_(parts, 1);
it = checkedToDouble_(parts, 2);
ch = checkedToInt_(parts, 3);
cf.setRT(rt);
cf.setMZ(mz);
cf.setIntensity(it);
if (input_type != TYPE_OLD_NOCHARGE)
cf.setCharge(ch);
}
catch (Exception::BaseException&)
{
throw Exception::ParseError(__FILE__, __LINE__, __PRETTY_FUNCTION__, "", String("Failed parsing in line ") + String((input_it - input.begin()) + 1) + ": Could not convert the first three columns to a number!\nOffending line: '" + line_trimmed + "' (line " + String((input_it - input.begin()) + 1) + ")\n");
}
// Check all features in one line
for (Size j = 1; j < input_features; ++j)
{
try
{
Feature f;
f.setUniqueId();
// Convert values. Will return -1 if not available.
rt = checkedToDouble_(parts, j * 4 + 0);
mz = checkedToDouble_(parts, j * 4 + 1);
it = checkedToDouble_(parts, j * 4 + 2);
ch = checkedToInt_(parts, j * 4 + 3);
// Only accept features with at least RT and MZ set
if (rt != -1 && mz != -1)
{
f.setRT(rt);
f.setMZ(mz);
f.setIntensity(it);
f.setCharge(ch);
cf.insert(j - 1, f);
}
}
catch (Exception::BaseException&)
{
throw Exception::ParseError(__FILE__, __LINE__, __PRETTY_FUNCTION__, "", String("Failed parsing in line ") + String((input_it - input.begin()) + 1) + ": Could not convert one of the four sub-feature columns (starting at column " + (j * 4 + 1) + ") to a number! Is the correct separator specified?\nOffending line: '" + line_trimmed + "' (line " + String((input_it - input.begin()) + 1) + ")\n");
}
}
//parse meta data
for (Size j = input_features * 4; j < parts.size(); ++j)
{
String part_trimmed = parts[j];
part_trimmed.trim();
if (part_trimmed != "")
{
//check if column name is ok
if (headers.size() <= j || headers[j] == "")
{
throw Exception::ParseError(__FILE__, __LINE__, __PRETTY_FUNCTION__, "",
String("Error: Missing meta data header for column ") + (j + 1) + "!"
+ String("Offending header line: '") + header_trimmed + "' (line 1)");
}
//add meta value
cf.setMetaValue(headers[j], part_trimmed);
}
}
//insert feature to map
consensus_map.push_back(cf);
}
// register FileDescriptions
ConsensusMap::FileDescription fd;
fd.filename = filename;
fd.size = consensus_map.size();
Size maps = std::max(input_features - 1, Size(1)); // its either a simple feature or a consensus map
// (in this case the 'input_features' includes the centroid, which we do not count)
for (Size i = 0; i < maps; ++i)
{
fd.label = String("EDTA_Map ") + String(i);
consensus_map.getFileDescriptions()[i] = fd;
}
}