public static IDataView LoadData(MLContext mlContext, String path)
{
DataTable dt = IdentificationFeatureExtractionUtil.loadDataTable(path);
IDataView dataView = IdentificationFeatureExtractionUtil.transformFeatures(dt, true);
//TrainTestData splitDataView = mlContext.Data.TrainTestSplit(dataView, testFraction: 0.2);
return(dataView);
}
//static string mzmlFileBaseName="MS_QC_240min";
public static void ExtractFeatures(String ms2File, out String extractedFeatureSavedFile_posAndNeg, out String extractedFeatureSavedFile_posAndNonPos)
{
Console.WriteLine("Extracting features from {0}", ms2File);
InputFileOrganizer.MS2SimulationTestFile = ms2File;
//InputFileOrganizer.MZMLSimulationTestFile = mzmlFile;
String ms2FileBaseName = Path.GetFileNameWithoutExtension(ms2File);
OutputFile_PositiveAndNegative = Path.Combine(InputFileOrganizer.OutputFolderOfTheRun, ms2FileBaseName + "_extractedFeatures_PositiveAndNegative.tsv");
OutputFile_PositiveAndNonPositive = Path.Combine(InputFileOrganizer.OutputFolderOfTheRun, ms2FileBaseName + "_extractedFeatures_positiveAndNonPositive.tsv");
OutputFile_PositiveAndNonPositive_NoDecoy = Path.Combine(InputFileOrganizer.OutputFolderOfTheRun, ms2FileBaseName + "_extractedFeatures_positiveAndNonPositive_NoDecoy.tsv");
//the current feature extraction will include decoy proteins in the database and testing set
SimulationWithDecoyParamsSetUp();
//placeholder values, dont matter
GlobalVar.ppmTolerance = 1;
GlobalVar.retentionTimeWindowSize = 1;
GlobalVar.AccordThreshold = 1;
GlobalVar.XCorr_Threshold = 1;
GlobalVar.NumDBThreshold = 1;
//
log.Info("Running No Exclusion Simulation");
ExclusionProfile exclusionProfile = ExclusionExplorer.SingleSimulationRun(ExclusionProfileEnum.NO_EXCLUSION_PROFILE);
log.Info("Extracting identification feature from exclusion profile");
List <IdentificationFeatures> idf = exclusionProfile.getFeatures();
log.Info("Recalibrating stDev");
idf = IdentificationFeatureExtractionUtil.recalibrateStDev(idf);
writeFeatures(idf);
extractedFeatureSavedFile_posAndNeg = OutputFile_PositiveAndNegative;
extractedFeatureSavedFile_posAndNonPos = OutputFile_PositiveAndNonPositive;
Console.WriteLine("Extracted Feature written to {0} and {1}", OutputFile_PositiveAndNegative, OutputFile_PositiveAndNonPositive);
}
protected void evaluateIdentification(IDs id)
{
// check if the peptide is identified or not
if (id == null)
{
performanceEvaluator.countMS2UnidentifiedAnalyzed();
return;
}
Peptide pep = getPeptideFromIdentification(id); // id is null, it already returned
// add decoy or non-existent protein connections
// database.AddProteinFromIdentification(pep, id.getParentProteinAccessions());
Double xCorr = id.getXCorr();
Double dCN = id.getDeltaCN();
pep.addScore(xCorr, dCN);
#if (!DONTEVALUATE)
performanceEvaluator.evaluateAnalysis(exclusionList, pep);
#endif
// exclude this peptide for analysis if the xCorr score is above a threshold
const double XCORR_THRESHOLD = 2.5;
// add the peptide to the exclusion list if it is over the xCorr threshold
if ((xCorr > XCORR_THRESHOLD))
{
performanceEvaluator.countPeptidesExcluded();
log.Debug("xCorrThreshold passed. Peptide added to the exclusion list.");
exclusionList.addPeptide(pep);
// calibrates our retention time alignment if the observed time is different
// from the predicted only if it passes this threshold
calibrateRetentionTime(pep);
}
// Add all the peptides corresponding to the parent protein, if the parent
// protein is deemed confidently identified by the logisitc regression
// classifier
Dictionary <String, Boolean> identificationPredictions = IdentificationFeatureExtractionUtil
.assessProteinIdentificationConfidence(pep.getProteins(), lrAccord);
List <Protein> proteinsToExclude = new List <Protein>();
foreach (Protein parentProtein in pep.getProteins())
{
// prevents repeated exclusion of a protein already excluded
if ((!parentProtein.IsExcluded()))
{
// determine if parent protein is confidently identified
bool isConfidentlyIdentified = identificationPredictions[parentProtein.getAccession()];
if (isConfidentlyIdentified)
{
// exclude all peptides of that protein
parentProtein.setExcluded(true);
log.Debug("Parent protein " + parentProtein.getAccession() + " is identified confidently "
+ parentProtein.getNumDB() + " times!");
performanceEvaluator.countProteinsExcluded();
proteinsToExclude.Add(parentProtein);
}
}
}
exclusionList.addProteins(proteinsToExclude);
}
protected void evaluateIdentification(IDs id)
{
// check if the peptide is identified or not
if (id == null)
{
performanceEvaluator.countMS2UnidentifiedAnalyzed();
return;
}
Peptide pep = getPeptideFromIdentification(id); // id is null, it already returned
//log.Info("Peptide Observed Time: {0}\tPredicted Time: {1} -----------------", id.getScanTime(),pep.getRetentionTime().getRetentionTimeStart());
// add decoy or non-existent protein connections
// database.AddProteinFromIdentification(pep, id.getParentProteinAccessions());
Double xCorr = id.getXCorr();
double dCN = id.getDeltaCN();
pep.addScore(xCorr, dCN);
#if (!DONTEVALUATE)
performanceEvaluator.evaluateAnalysis(exclusionList, pep);
#endif
//RetentionTime rt = pep.getRetentionTime();
//if (!rtCalcPredictedRT.Keys.Contains(pep.getSequence()))
//{
// rtCalcPredictedRT.Add(pep.getSequence(), rt.getRetentionTimePeak());
//}
//double[] values = new double[] { id.getScanTime(), id.getXCorr(), rt.getRetentionTimePeak(), rt.getRetentionTimeStart() + GlobalVar.retentionTimeWindowSize, RetentionTime.getRetentionTimeOffset(), rtCalcPredictedRT[pep.getSequence()], rt.IsPredicted() ? 1 : 0 };
// exclude this peptide for analysis if the xCorr score is above a threshold
const double XCORR_THRESHOLD = 2.5;
// add the peptide to the exclusion list if it is over the xCorr threshold
if ((xCorr > XCORR_THRESHOLD))
{
performanceEvaluator.countPeptidesExcluded();
log.Debug("xCorrThreshold passed. Peptide added to the exclusion list.");
exclusionList.addPeptide(pep);
// calibrates our retention time alignment if the observed time is different
// from the predicted only if it passes this threshold
calibrateRetentionTime(pep);
}
// Add all the peptides corresponding to the parent protein, if the parent
// protein is deemed confidently identified by the logisitc regression
// classifier
Dictionary <String, Boolean> identificationPredictions = IdentificationFeatureExtractionUtil
.assessProteinIdentificationConfidence(pep.getProteins(), lrAccord);
List <Protein> proteinsToExclude = new List <Protein>();
foreach (Protein parentProtein in pep.getProteins())
{
// prevents repeated exclusion of a protein already excluded
if ((!parentProtein.IsExcluded()))
{
// determine if parent protein is confidently identified
bool isConfidentlyIdentified = identificationPredictions[parentProtein.getAccession()];
if (isConfidentlyIdentified)
{
// exclude all peptides of that protein
#if TRACKEXCLUDEDPROTEINFEATURE
excludedProteinFeatureList.Add(parentProtein.vectorize().ItemArray);
#endif
parentProtein.setExcluded(true);
log.Debug("Parent protein " + parentProtein.getAccession() + " is identified confidently "
+ parentProtein.getNumDB() + " times!");
performanceEvaluator.countProteinsExcluded();
proteinsToExclude.Add(parentProtein);
}
}
}
exclusionList.addProteins(proteinsToExclude);
}
public DataRow vectorize()
{
return(IdentificationFeatureExtractionUtil.extractFeatureVector(accession, peptideScores));
}
public IdentificationFeatures extractFeatures()
{
return(IdentificationFeatureExtractionUtil.extractFeatures(accession, peptideScores));
}
