public static MassCalibrationResults RunMassCalibration(IEnumerable <Lipid> lipidList, LcMsRun lcmsRun, double hcdMassError, IProgress <int> progress = null)
{
var ppmErrorList = new List <double>();
var lipidsGroupedByTarget = lipidList.OrderBy(x => x.LipidTarget.Composition.Mass).GroupBy(x => x.LipidTarget).ToList();
foreach (var kvp in lipidsGroupedByTarget)
{
var target = kvp.Key;
var spectrumSearchResultList = InformedWorkflow.RunInformedWorkflow(target, lcmsRun, hcdMassError, 500, scoreModel: null);
if (spectrumSearchResultList.Any())
{
var targetIon = new Ion(target.Composition - Composition.Hydrogen, 1);
var targetMz = targetIon.GetMonoIsotopicMz();
var bestSpectrumSearchResult = spectrumSearchResultList.OrderBy(x => x.Score).First();
var massSpectrum = bestSpectrumSearchResult.PrecursorSpectrum.Peaks;
var closestPeak = massSpectrum.OrderBy(x => Math.Abs(x.Mz - targetMz)).First();
var ppmError = LipidUtil.PpmError(targetMz, closestPeak.Mz);
ppmErrorList.Add(ppmError);
}
}
var ppmHistogram = QcUtil.CalculateHistogram(ppmErrorList, hcdMassError, 0.25);
var massCalibrationResults = new MassCalibrationResults(ppmHistogram);
return(massCalibrationResults);
}
private async void ProcessAllTargetsButtonClick(object sender, RoutedEventArgs e)
{
var fragmentationMode = (FragmentationMode)FragmentationModeComboBox.SelectedItem;
var precursorError = double.Parse(PrecursorErrorTextBox.Text);
var hcdMassError = double.Parse(HcdErrorTextBox.Text);
var cidMassError = double.Parse(CidErrorTextBox.Text);
var resultsPerScan = int.Parse(ResultsPerScanTextBox.Text);
LipidGroupSearchResultsDataGrid.Visibility = Visibility.Hidden;
ExportGlobalResultsButton.Visibility = Visibility.Hidden;
ExportAllGlobalResultsButton.Visibility = Visibility.Hidden;
await Task.Run(() => SingleTargetViewModel.OnProcessAllTarget(hcdMassError, cidMassError, fragmentationMode, resultsPerScan));
LipidGroupSearchResultsDataGrid.Visibility = Visibility.Visible;
ExportGlobalResultsButton.Visibility = Visibility.Visible;
ExportAllGlobalResultsButton.Visibility = Visibility.Visible;
// Select the best spectrum search result
if (SingleTargetViewModel.LipidGroupSearchResultList.Count > 0)
{
double currentMz = 0;
List <LipidGroupSearchResult> removeList = new List <LipidGroupSearchResult>();
foreach (LipidGroupSearchResult entry in SingleTargetViewModel.LipidGroupSearchResultList)
{
var targetMz = entry.LipidTarget.MzRounded;
if (entry.SpectrumSearchResult.PrecursorSpectrum != null)
{
var massSpectrum = entry.SpectrumSearchResult.PrecursorSpectrum.Peaks;
var closestPeak = massSpectrum.OrderBy(x => Math.Abs(x.Mz - targetMz)).First();
currentMz = closestPeak.Mz;
}
else
{
var isolationMz = entry.SpectrumSearchResult.HcdSpectrum?.IsolationWindow.IsolationWindowTargetMz ??
entry.SpectrumSearchResult.CidSpectrum.IsolationWindow.IsolationWindowTargetMz;
currentMz = isolationMz;
}
double currentPpmError = LipidUtil.PpmError(targetMz, currentMz);
if (Math.Abs(currentPpmError) > Math.Abs(precursorError))
{
removeList.Add(entry);
}
}
foreach (LipidGroupSearchResult remove in removeList)
{
SingleTargetViewModel.LipidGroupSearchResultList.Remove(remove);
}
var dataGrid = LipidGroupSearchResultsDataGrid;
dataGrid.SelectedItem = SingleTargetViewModel.LipidGroupSearchResultList[0];
dataGrid.ScrollIntoView(SingleTargetViewModel.LipidGroupSearchResultList[0]);
}
}
private static void OutputResultsToTsv(
IReadOnlyCollection <LipidGroupSearchResult> lipidGroupSearchResults,
string fileLocation,
string rawFileName,
IProgress <int> progress = null,
bool writeHeader = true)
{
using (TextWriter textWriter = new StreamWriter(fileLocation, true))
{
if (writeHeader)
{
textWriter.WriteLine("Raw Data File\tLM_ID\tCommon Name\tAdduct\tCategory\tMain Class\tSub Class\tExact m/z\tFormula\tObserved m/z\tppm Error\tApex RT\tPrecursor RT\tApex NET\tIntensity\tPeak Area\tScore\tMS/MS Scan\tPrecursor Scan\tApex Scan\tPUBCHEM_SID\tPUBCHEM_CID\tINCHI_KEY\tKEGG_ID\tHMDBID\tCHEBI_ID\tLIPIDAT_ID\tLIPIDBANK_ID");
}
var progressCounter = 0;
foreach (var lipidGroupSearchResult in lipidGroupSearchResults)
{
var lipidTarget = lipidGroupSearchResult.LipidTarget;
var spectrumSearchResult = lipidGroupSearchResult.SpectrumSearchResult;
var Precursor = spectrumSearchResult.PrecursorSpectrum != null;
var targetMz = lipidTarget.MzRounded;
var observedMz = spectrumSearchResult.HcdSpectrum?.IsolationWindow.IsolationWindowTargetMz ??
spectrumSearchResult.CidSpectrum.IsolationWindow.IsolationWindowTargetMz;
if (Precursor)
{
var massSpectrum = spectrumSearchResult.PrecursorSpectrum.Peaks;
var closestPeak = massSpectrum.OrderBy(x => Math.Abs(x.Mz - targetMz)).First();
observedMz = closestPeak.Mz;
}
var score = lipidGroupSearchResult.Score;
var msmsScan = spectrumSearchResult.HcdSpectrum?.ScanNum ?? spectrumSearchResult.CidSpectrum.ScanNum;
var ppmError = LipidUtil.PpmError(targetMz, observedMz);
foreach (var lipid in lipidGroupSearchResult.LipidList)
{
var line = new StringBuilder();
line.Append(rawFileName + "\t");
line.Append(lipid.LipidMapsId + "\t");
line.Append(lipidTarget.StrippedDisplay + "\t");
line.Append(lipid.AdductFull + "\t");
line.Append(lipid.Category + "\t");
line.Append(lipid.MainClass + "\t");
line.Append(lipid.SubClass + "\t");
line.Append(lipidTarget.MzRounded + "\t");
line.Append(lipidTarget.EmpiricalFormula + "\t");
line.Append(observedMz + "\t");
line.Append(ppmError + "\t");
line.Append(spectrumSearchResult.RetentionTime + "\t");
if (Precursor)
{
line.Append(spectrumSearchResult.PrecursorSpectrum.ElutionTime + "\t");
}
else
{
line.Append("\t");
}
line.Append(spectrumSearchResult.NormalizedElutionTime + "\t");
line.Append(spectrumSearchResult.ApexIntensity + "\t");
line.Append(spectrumSearchResult.PeakArea + "\t");
line.Append(score + "\t");
line.Append(msmsScan + "\t");
if (Precursor)
{
line.Append(spectrumSearchResult.PrecursorSpectrum.ScanNum + "\t");
}
else
{
line.Append("\t");
}
line.Append(spectrumSearchResult.ApexScanNum + "\t");
line.Append(lipid.PubChemSid + "\t");
line.Append(lipid.PubChemCid + "\t");
line.Append(lipid.InchiKey + "\t");
line.Append(lipid.KeggId + "\t");
line.Append(lipid.HmdbId + "\t");
line.Append(lipid.ChebiId + "\t");
line.Append(lipid.LipidatId + "\t");
line.Append(lipid.LipidBankId + "\t");
textWriter.WriteLine(line.ToString());
}
if (progress != null)
{
progressCounter++;
var currentProgress = (int)(progressCounter / (double)lipidGroupSearchResults.Count * 100);
progress.Report(currentProgress);
}
}
}
}
private static void OutputResultsToMzTab(IReadOnlyCollection <LipidGroupSearchResult> lipidGroupSearchResults,
string fileLocation, string rawFileName, IProgress <int> progress = null)
{
using (TextWriter textWriter = new StreamWriter(fileLocation))
{
var progressCounter = 0;
var mods = new List <string>();
foreach (var lipidGroupSearchResult in lipidGroupSearchResults)
{
foreach (var lipid in lipidGroupSearchResult.LipidList)
{
if (!mods.Contains(lipid.AdductFull))
{
mods.Add(lipid.AdductFull);
}
}
}
//Write meta-data
textWriter.WriteLine("MTD\tmzTabVersion\t1.0 rc5");
textWriter.WriteLine("MTD\tmzTab-mode\tComplete");
textWriter.WriteLine("MTD\tmzTab-type\tQuantification");
textWriter.WriteLine("MTD\tsoftware[1]\t[, , LIQUID, ]");
textWriter.WriteLine("MTD\tsmallmolecule_search_engine_score[1]\t[, , LIQUID_Score_Analyzer, ]");
textWriter.WriteLine("MTD\tfixed_mod[1]\t[MS, MS:1002038, unlabeled sample, ]");
foreach (var variableMod in mods)
{
textWriter.WriteLine("MTD\tvariable_mod[1]\t[, , {0}]", variableMod);
}
textWriter.WriteLine("MTD\tquantification_method\t[, , LIQUID_Analysis, ]");
textWriter.WriteLine("MTD\tsmall_molecule-quantification_unit\t[PRIDE, PRIDE:0000330, Arbitrary quantification unit, ]");
//Get the raw/mzml location
textWriter.WriteLine("MTD\tms_run[1]-location\t{0}", rawFileName); //TODO:
textWriter.WriteLine("MTD\tassay[1]-quantification_reagent\t[MS, MS:1002038, unlabeled sample, ]");
textWriter.WriteLine("MTD\tassay[1]-ms_run_ref\tms_run[1]");
textWriter.WriteLine("MTD\tstudy_variable[1]-assay_refs\tassay[1]");
textWriter.WriteLine("MTD\tstudy_variable[1]-description\tLIQUID Quantification");
textWriter.WriteLine("MTD\tcolunit-small_molecule\tretention_time=[UO, UO:0000031, minute, ]");
textWriter.WriteLine("");
//Write small molecule section headers
textWriter.WriteLine("SMH\tidentifier\tchemical_formula\tsmiles\tinchi_key\tdescription\texp_mass_to_charge\tcalc_mass_to_charge\tcharge\tretention_time\ttaxid\tspecies\tdatabase\tdatabase_version\tspectra_ref\tsearch_engine\tbest_search_engine_score[1]\tsearch_engine_score[1]_ms_run[1]\tmodification\tsmallmolecule_abundance_assay[1]\tsmallmolecule_abundance_study_variable[1]\tsmallmolecule_abundance_stdev_study_variable[1]\tsmallmolecule_abundance_std_error_study_variable[1]");
//Write small molecule section datas
foreach (var lipidGroupSearchResult in lipidGroupSearchResults)
{
var lipidTarget = lipidGroupSearchResult.LipidTarget;
var spectrumSearchResult = lipidGroupSearchResult.SpectrumSearchResult;
var targetMz = lipidTarget.MzRounded;
var massSpectrum = spectrumSearchResult.PrecursorSpectrum.Peaks;
var closestPeak = massSpectrum.OrderBy(x => Math.Abs(x.Mz - targetMz)).First();
var observedMz = closestPeak.Mz;
var ppmError = LipidUtil.PpmError(targetMz, closestPeak.Mz);
var score = lipidGroupSearchResult.Score;
var msmsScan = spectrumSearchResult.HcdSpectrum?.ScanNum ?? spectrumSearchResult.CidSpectrum.ScanNum;
//TODO: var charge = calculatedFromMZ
foreach (var lipid in lipidGroupSearchResult.LipidList)
{
var line = new StringBuilder();
line.Append("SML\t");
//var indexToStartRemove = lipid.LipidTarget.StrippedDisplay.IndexOf("/");
//var id = lipid.LipidTarget.StrippedDisplay.Substring(0, indexToStartRemove);
//id = id.Replace("(", "");
var id = lipid.LipidTarget.StrippedDisplay;
line.Append(id + "\t");
//if (!string.IsNullOrWhiteSpace(lipid.LipidMapsId))
// line.Append(lipid.LipidMapsId + "\t"); // identifier
//else
// line.Append("null" + "\t");
line.Append(lipidTarget.EmpiricalFormula + "\t"); // chemical_formula
line.Append("null" + "\t"); // smiles
if (!string.IsNullOrWhiteSpace(lipid.InchiKey))
{
line.Append(lipid.InchiKey + "\t"); // inchi_key
}
else
{
line.Append("null" + "\t");
}
line.Append(lipid.SubClass + " : " + lipidTarget.FragmentationMode.ToString() + " charge" + "\t"); // description
line.Append(observedMz + "\t"); // exp_mass_to_charge
line.Append(lipidTarget.MzRounded + "\t"); // calc_mass_to_charge
line.Append("1" + "\t"); // charge
line.Append(spectrumSearchResult.RetentionTime + "\t"); // retention_time
line.Append("null" + "\t"); // taxid
line.Append("null" + "\t"); // species
line.Append("null" + "\t"); // database
line.Append("null" + "\t"); // database_version
//line.Append(spectrumSearchResult.PrecursorSpectrum.ScanNum + "\t"); // spectra_ref
line.Append(msmsScan + "\t"); // spectra_ref
line.Append("[, , Liquid, ]" + "\t"); // search_engine
line.Append(score + "\t"); // best_search_engine_score[1]
line.Append(score + "\t"); // search_engine_score[1]_ms_run[1]
line.Append(lipid.AdductFull + "\t"); // FROM ADDUCTFULL
line.Append(spectrumSearchResult.ApexIntensity + "\t"); // small_molecule_abundance_assay[1]
line.Append("null" + "\t"); // ^^Study_variable[1]
line.Append("null" + "\t"); // stdev_study_variable[1]
line.Append("null" + "\t"); // std_err_study_variable[1]
textWriter.WriteLine(line.ToString());
}
if (progress != null)
{
progressCounter++;
var currentProgress = (int)(progressCounter / (double)lipidGroupSearchResults.Count * 100);
progress.Report(currentProgress);
}
}
}
}
public static List <SpectrumSearchResult> RunInformedWorkflow(
LipidTarget target,
LcMsRun lcmsRun,
double hcdMassError,
double cidMassError,
ScoreModel scoreModel = null)
{
IEnumerable <MsMsSearchUnit> msMsSearchUnits = target.GetMsMsSearchUnits();
// I have to subtract an H for the target Ion since InformedProteomics will assume protenated
var targetIon = new Ion(target.Composition - Composition.Hydrogen, 1);
var targetMz = target.MzRounded;
var hcdTolerance = new Tolerance(hcdMassError, ToleranceUnit.Ppm);
var cidTolerance = new Tolerance(cidMassError, ToleranceUnit.Ppm);
var activationMethodCombination = GlobalWorkflow.FigureOutActivationMethodCombination(lcmsRun);
// Find out which MS/MS scans have a precursor m/z that matches the target
//List<int> matchingMsMsScanNumbers = lcmsRun.GetFragmentationSpectraScanNums(targetIon).ToList();
var matchingMsMsScanNumbers = lcmsRun.GetFragmentationSpectraScanNums(targetMz).ToList();
var spectrumSearchResultList = new List <SpectrumSearchResult>();
for (var i = 0; i + 1 < matchingMsMsScanNumbers.Count; i += 2)
{
var firstScanNumber = matchingMsMsScanNumbers[i];
var secondScanNumber = matchingMsMsScanNumbers[i + 1];
// Scan numbers should be consecutive.
/*
* if (secondScanNumber - firstScanNumber != 1)
* {
* i--;
* continue;
* }
* */
ProductSpectrum firstMsMsSpectrum = null;
ProductSpectrum secondMsMsSpectrum = null;
if (activationMethodCombination == ActivationMethodCombination.CidThenHcd ||
activationMethodCombination == ActivationMethodCombination.HcdThenCid)
{
// Lookup the MS/MS Spectrum
firstMsMsSpectrum = lcmsRun.GetSpectrum(firstScanNumber) as ProductSpectrum;
if (firstMsMsSpectrum == null)
{
continue;
}
// Lookup the MS/MS Spectrum
secondMsMsSpectrum = lcmsRun.GetSpectrum(secondScanNumber) as ProductSpectrum;
if (secondMsMsSpectrum == null)
{
continue;
}
}
else if (activationMethodCombination == ActivationMethodCombination.CidOnly ||
activationMethodCombination == ActivationMethodCombination.HcdOnly)
{
firstMsMsSpectrum = lcmsRun.GetSpectrum(firstScanNumber) as ProductSpectrum;
}
if (firstMsMsSpectrum == null)
{
continue;
}
// Filter MS/MS Spectrum based on mass error
var msMsPrecursorMz = firstMsMsSpectrum.IsolationWindow.IsolationWindowTargetMz;
//if (Math.Abs(msMsPrecursorMz - targetMz) > 0.4) continue;
var ppmError = LipidUtil.PpmError(targetMz, msMsPrecursorMz);
if (Math.Abs(ppmError) > hcdMassError)
{
continue;
}
// Assign each MS/MS spectrum to HCD or CID
ProductSpectrum hcdSpectrum;
ProductSpectrum cidSpectrum;
if (firstMsMsSpectrum.ActivationMethod == ActivationMethod.HCD)
{
hcdSpectrum = firstMsMsSpectrum;
cidSpectrum = secondMsMsSpectrum;
}
else
{
hcdSpectrum = secondMsMsSpectrum;
cidSpectrum = firstMsMsSpectrum;
}
// Get all matching peaks
var hcdSearchResultList = hcdSpectrum != null ? (from msMsSearchUnit in msMsSearchUnits let peak = hcdSpectrum.FindPeak(msMsSearchUnit.Mz, hcdTolerance) select new MsMsSearchResult(msMsSearchUnit, peak)).ToList() : new List <MsMsSearchResult>();
var cidSearchResultList = cidSpectrum != null ? (from msMsSearchUnit in msMsSearchUnits let peak = cidSpectrum.FindPeak(msMsSearchUnit.Mz, cidTolerance) select new MsMsSearchResult(msMsSearchUnit, peak)).ToList() : new List <MsMsSearchResult>();
// Find the MS1 data
//Xic xic = lcmsRun.GetPrecursorExtractedIonChromatogram(targetMz, hcdTolerance, firstScanNumber);
var precursorScanNumber = 0;
if (lcmsRun.MinMsLevel == 1) //Make sure there are precursor scans in file
{
precursorScanNumber = lcmsRun.GetPrecursorScanNum(matchingMsMsScanNumbers[i]);
}
var precursorSpectrum = lcmsRun.GetSpectrum(precursorScanNumber);
var xic = lcmsRun.GetFullPrecursorIonExtractedIonChromatogram(targetMz, hcdTolerance);
// Bogus data
if (precursorSpectrum != null && (xic.GetApexScanNum() < 0 || xic.GetSumIntensities() <= 0))
{
continue;
}
SpectrumSearchResult spectrumSearchResult;
if (precursorSpectrum != null)
{
spectrumSearchResult = new SpectrumSearchResult(hcdSpectrum, cidSpectrum, precursorSpectrum, hcdSearchResultList, cidSearchResultList, xic, lcmsRun, scoreModel, target)
{
PrecursorTolerance = new Tolerance(hcdMassError, ToleranceUnit.Ppm)
};
}
else //If there are no precursor scans in this file
{
spectrumSearchResult = new SpectrumSearchResult(hcdSpectrum, cidSpectrum, hcdSearchResultList, cidSearchResultList, lcmsRun, scoreModel, target)
{
PrecursorTolerance = new Tolerance(hcdMassError, ToleranceUnit.Ppm),
};
}
spectrumSearchResultList.Add(spectrumSearchResult);
}
return(spectrumSearchResultList);
}