private void GetChromatogramCorrelationData(SipperLcmsTargetedResult result)
{
_chromScanWindowWidth = result.ChromPeakSelected.Width * 2;
var startScan = result.ScanSet.PrimaryScanNumber - (int)Math.Round(_chromScanWindowWidth / 2, 0);
var stopScan = result.ScanSet.PrimaryScanNumber + (int)Math.Round(_chromScanWindowWidth / 2, 0);
result.ChromCorrelationData = _chromatogramCorrelatorTask.CorrelatePeaksWithinIsotopicProfile(result.Run,
NormalizedIso,
startScan,
stopScan);
ChromatogramRSquaredVals.AddRange(
result.ChromCorrelationData.CorrelationDataItems.Select(p => p.CorrelationRSquaredVal.GetValueOrDefault(-1)).
ToList());
//trim off zeros
for (var i = ChromatogramRSquaredVals.Count - 1; i >= 0; i--)
{
if (ChromatogramRSquaredVals[i] <= 0)
{
ChromatogramRSquaredVals.RemoveAt(i);
}
else
{
break;
}
}
result.ChromCorrelationMin = ChromatogramRSquaredVals.Min();
if (ChromatogramRSquaredVals.Count > 1)
{
//get the best rsquared value other than the base peak's rsquared value (which is always 1)
result.ChromCorrelationMax =
(from n in ChromatogramRSquaredVals orderby n descending select n).ToList().ElementAt(1);
result.ChromCorrelationMedian = MathUtils.GetMedian(ChromatogramRSquaredVals);
result.ChromCorrelationAverage = ChromatogramRSquaredVals.Average();
if (ChromatogramRSquaredVals.Count > 2)
{
result.ChromCorrelationStdev = MathUtils.GetStDev(ChromatogramRSquaredVals);
}
else
{
result.ChromCorrelationStdev = -1;
}
}
else
{
result.ChromCorrelationMax = 1;
}
}
private void GetChromatogramCorrelationData(SipperLcmsTargetedResult result)
{
//NOTE: this is a very important step. If too wide, correlation data will be poor
//and may lead to overly strict treatment of the data
_chromScanWindowWidth = result.ChromPeakSelected.Width / 2.35 * 4;
//TODO: change this!
//_chromScanWindowWidth = 19;
var startScan = result.ScanSet.PrimaryScanNumber - (int)Math.Round(_chromScanWindowWidth / 2, 0);
var stopScan = result.ScanSet.PrimaryScanNumber + (int)Math.Round(_chromScanWindowWidth / 2, 0);
result.ChromCorrelationData = _chromatogramCorrelatorTask.CorrelatePeaksWithinIsotopicProfile(result.Run, NormalizedIso, startScan, stopScan);
if (result.ChromCorrelationData.CorrelationDataItems.Count > 0)
{
ChromatogramRSquaredVals.AddRange(result.ChromCorrelationData.CorrelationDataItems.Select(p => p.CorrelationRSquaredVal.GetValueOrDefault(-1)).ToList());
}
//GORD: remove this console code later
//Console.WriteLine();
//Console.WriteLine(result.Target.ID);
//foreach (var item in result.ChromCorrelationData.CorrelationDataItems)
//{
// Console.WriteLine(item.CorrelationRSquaredVal);
//}
//trim off zeros
for (var i = ChromatogramRSquaredVals.Count - 1; i >= 1; i--)
{
if (ChromatogramRSquaredVals[i] <= 0)
{
ChromatogramRSquaredVals.RemoveAt(i);
}
else
{
break;
}
}
if (ChromatogramRSquaredVals.Count > 1)
{
result.ChromCorrelationMin = ChromatogramRSquaredVals.Min();
//get the best rsquared value other than the base peak's rsquared value (which is always 1)
result.ChromCorrelationMax =
(from n in ChromatogramRSquaredVals orderby n descending select n).ToList().ElementAt(1);
result.ChromCorrelationMedian = MathUtils.GetMedian(ChromatogramRSquaredVals);
result.ChromCorrelationAverage = ChromatogramRSquaredVals.Average();
if (ChromatogramRSquaredVals.Count > 2)
{
result.ChromCorrelationStdev = MathUtils.GetStDev(ChromatogramRSquaredVals);
}
else
{
result.ChromCorrelationStdev = -1;
}
}
else
{
result.ChromCorrelationMin = -1;
result.ChromCorrelationMax = -1;
}
}
public override void Execute(ResultCollection resultList)
{
Check.Require(resultList.CurrentTargetedResult is SipperLcmsTargetedResult, "Sipper Quantifier only works on Sipper-type result objects");
Check.Require(resultList.Run.CurrentMassTag != null, this.Name + " failed; CurrentMassTag is empty");
Check.Require(resultList.Run.CurrentMassTag.IsotopicProfile != null, this.Name + " failed; Theor isotopic profile is empty. Run a TheorFeatureGenerator");
var result = (SipperLcmsTargetedResult)resultList.CurrentTargetedResult;
result.AreaUnderDifferenceCurve = -9999;
result.AreaUnderRatioCurve = -9999;
RatioVals.Xvalues = new double[] { 1, 2, 3, 4, 5, 6, 7 };
RatioVals.Yvalues = new double[] { 0, 0, 0, 0, 0, 0, 0 };
RatioLogVals.Xvalues = new double[] { 1, 2, 3, 4, 5, 6, 7 };
RatioLogVals.Yvalues = new double[] { 0, 0, 0, 0, 0, 0, 0 };
ChromatogramRSquaredVals.Clear();
Check.Require(result != null, "No MassTagResult has been generated for CurrentMassTag");
if (result.IsotopicProfile == null || result.IsotopicProfile.Peaklist == null || result.IsotopicProfile.Peaklist.Count < 2)
{
return;
}
var theorUnlabelledIso = resultList.Run.CurrentMassTag.IsotopicProfile.CloneIsotopicProfile();
IsotopicProfileUtilities.NormalizeIsotopicProfile(theorUnlabelledIso);
//PeakUtilities.TrimIsotopicProfile(unlabeledIso, 0.001);
var sb = new StringBuilder();
sb.Append(result.Target.ID + "\t" + result.Target.MZ.ToString("0.0000") + "\t" + result.Target.ChargeState +
"-----------------------------\n");
var indexMostAbundantTheorPeak = theorUnlabelledIso.GetIndexOfMostIntensePeak();
NormalizedIso = result.IsotopicProfile.CloneIsotopicProfile();
IsotopicProfileUtilities.NormalizeIsotopicProfileToSpecificPeak(NormalizedIso, indexMostAbundantTheorPeak);
// return;
if (result.Flags.Count > 0)
{
var flagstring = "";
foreach (var resultFlag in result.Flags)
{
flagstring += resultFlag.Description + ";";
}
result.ErrorDescription = flagstring.TrimEnd(';');
result.FailedResult = true;
if (flagstring.Contains("PeakToTheLeft"))
{
result.FailureType = Globals.TargetedResultFailureType.DeisotopingProblemDetected;
}
else
{
result.FailureType = Globals.TargetedResultFailureType.QuantifierFailure;
}
}
var resultPassesMinimalCriteria = (result.Score < MaximumFitScoreForFurtherProcessing && result.Flags.Count == 0);
if (resultPassesMinimalCriteria)
{
//------------------------------------- Get chromatogramCorrelation data -------------------------------
if (IsChromatogramCorrelationPerformed)
{
GetChromatogramCorrelationData(result);
}
NormalizedAdjustedIso = NormalizedIso.CloneIsotopicProfile();
//this is experimental!! it attempts to correct the problems caused by Orbitraps on isotopic profile intensities
//UpdateIsoIntensitiesUsingChromCorrData(result.ChromCorrelationData, NormalizedAdjustedIso);
//----------------- create ratio data -------------------------------------------------------
var ratioData = NormalizedAdjustedIso.CloneIsotopicProfile();
for (var i = 0; i < NormalizedIso.Peaklist.Count; i++)
{
if (i < theorUnlabelledIso.Peaklist.Count && theorUnlabelledIso.Peaklist[i].Height > MinimumRelativeIntensityForRatioCalc)
{
ratioData.Peaklist[i].Height = (NormalizedIso.Peaklist[i].Height / theorUnlabelledIso.Peaklist[i].Height - 1);
}
else
{
ratioData.Peaklist[i].Height = 0;
}
}
//trim off zeros from ratio data
for (var i = ratioData.Peaklist.Count - 1; i >= 0; i--)
{
if (ratioData.Peaklist[i].Height == 0)
{
ratioData.Peaklist.RemoveAt(i);
}
else
{
break;
}
}
var xvals = ratioData.Peaklist.Select((p, i) => new { peak = p, index = i }).Select(n => (double)n.index).ToList();
var yvals = ratioData.Peaklist.Select(p => (double)p.Height).ToList();
result.AreaUnderRatioCurve = yvals.Sum();
RatioVals.Xvalues = xvals.ToArray();
RatioVals.Yvalues = yvals.ToArray();
GetLinearRegressionData(result, xvals, yvals);
//------------- subtract unlabelled profile from normalized profile ----------------------
var subtractedIsoData = NormalizedAdjustedIso.CloneIsotopicProfile();
for (var i = 0; i < subtractedIsoData.Peaklist.Count; i++)
{
float intensityTheorPeak = 0;
if (i < theorUnlabelledIso.Peaklist.Count)
{
intensityTheorPeak = theorUnlabelledIso.Peaklist[i].Height;
}
var subtractedIntensity = subtractedIsoData.Peaklist[i].Height - intensityTheorPeak;
if (subtractedIntensity < 0)
{
subtractedIntensity = 0;
}
subtractedIsoData.Peaklist[i].Height = subtractedIntensity;
}
result.AreaUnderDifferenceCurve = subtractedIsoData.Peaklist.Select(p => p.Height).Sum();
result.NumCarbonsLabelled = calculateNumCarbonsFromSubtractedProfile(subtractedIsoData);
//-------------- calculate Label Distribution ------------------------------------------------
double[] numLabelVals;
double[] labelDistributionVals;
var theorIntensityVals = theorUnlabelledIso.Peaklist.Select(p => (double)p.Height).ToList();
var normalizedCorrectedIntensityVals = NormalizedAdjustedIso.Peaklist.Select(p => (double)p.Height).ToList();
var numRightPads = 3;
_labelingDistributionCalculator.CalculateLabelingDistribution(theorIntensityVals, normalizedCorrectedIntensityVals,
LabeldistCalcIntensityThreshold,
LabeldistCalcIntensityThreshold,
out numLabelVals,
out labelDistributionVals, true, true, 0, numRightPads, 0, 0);
//negative distribution values are zeroed out. And, then the remaining values are adjusted such that they add up to 1.
result.LabelDistributionVals = AdjustLabelDistributionVals(labelDistributionVals);
double distFractionUnlabelled, distFractionLabelled, distAverageLabelsIncorporated;
_labelingDistributionCalculator.OutputLabelingInfo(result.LabelDistributionVals, out distFractionUnlabelled,
out distFractionLabelled,
out distAverageLabelsIncorporated);
//-------------- make calculations using inputs from chrom correlation data -------------------
HighQualitySubtractedProfile = GetIsoDataPassingChromCorrelation(result.ChromCorrelationData, subtractedIsoData);
var highQualityRatioProfileData = GetIsoDataPassingChromCorrelation(result.ChromCorrelationData, ratioData);
if (highQualityRatioProfileData.Peaklist != null && highQualityRatioProfileData.Peaklist.Count > 0)
{
result.AreaUnderDifferenceCurve = HighQualitySubtractedProfile.Peaklist.Select(p => p.Height).Sum();
}
else
{
result.AreaUnderDifferenceCurve = 0;
}
if (HighQualitySubtractedProfile.Peaklist != null && HighQualitySubtractedProfile.Peaklist.Count > 0)
{
result.AreaUnderRatioCurveRevised = highQualityRatioProfileData.Peaklist.Select(p => p.Height).Sum();
}
else
{
result.AreaUnderRatioCurveRevised = 0;
}
var numCarbonsLabelled = GetNumCarbonsLabelledUsingAverageMassDifferences(theorUnlabelledIso,
HighQualitySubtractedProfile);
//result.NumCarbonsLabelled = numCarbonsLabelled;
result.NumCarbonsLabelled = distAverageLabelsIncorporated;
var numCarbons = result.Target.GetAtomCountForElement("C");
result.PercentCarbonsLabelled = (result.NumCarbonsLabelled / numCarbons) * 100;
result.PercentPeptideLabelled = distFractionLabelled * 100;
//if (HighQualitySubtractedProfile.Peaklist != null && HighQualitySubtractedProfile.Peaklist.Count > 0)
//{
// result.PercentPeptideLabelled = HighQualitySubtractedProfile.Peaklist.Max(p => p.Height) * 100;
//}
//else
//{
// result.PercentPeptideLabelled = 0;
//}
result.NumHighQualityProfilePeaks = highQualityRatioProfileData.Peaklist != null
? highQualityRatioProfileData.Peaklist.Count
: 0;
//Console.WriteLine();
//Console.WriteLine(result);
//foreach (var val in ChromatogramRSquaredVals)
//{
// Console.WriteLine(val);
//}
}
else
{
result.FailedResult = true;
if (result.FailureType == Globals.TargetedResultFailureType.None)
{
result.FailureType = Globals.TargetedResultFailureType.MinimalCriteriaNotMet;
}
}
}
public override void Execute(ResultCollection resultList)
{
Check.Require(resultList.CurrentTargetedResult is SipperLcmsTargetedResult, "Sipper Quantifier only works on Sipper-type result objects");
Check.Require(resultList.Run.CurrentMassTag != null, this.Name + " failed; CurrentMassTag is empty");
Check.Require(resultList.Run.CurrentMassTag.IsotopicProfile != null, this.Name + " failed; Theor isotopic profile is empty. Run a TheorFeatureGenerator");
ResetQuantifierData();
var result = (SipperLcmsTargetedResult)resultList.CurrentTargetedResult;
result.AreaUnderDifferenceCurve = -9999;
result.AreaUnderRatioCurve = -9999;
RatioVals.Xvalues = new double[] { 1, 2, 3, 4, 5, 6, 7 };
RatioVals.Yvalues = new double[] { 0, 0, 0, 0, 0, 0, 0 };
ChromatogramRSquaredVals.Clear();
FitScoreData.Clear();
Check.Require(result != null, "No MassTagResult has been generated for CurrentMassTag");
if (result.IsotopicProfile == null || result.IsotopicProfile.Peaklist == null || result.IsotopicProfile.Peaklist.Count < 2)
{
return;
}
var theorUnlabelledIso = resultList.Run.CurrentMassTag.IsotopicProfile.CloneIsotopicProfile();
IsotopicProfileUtilities.NormalizeIsotopicProfile(theorUnlabelledIso);
//PeakUtilities.TrimIsotopicProfile(unlabeledIso, 0.001);
var indexOfCorrespondingObservedPeak = PeakUtilities.getIndexOfClosestValue(result.IsotopicProfile.Peaklist,
theorUnlabelledIso.getMostIntensePeak().XValue, 0, result.IsotopicProfile.Peaklist.Count - 1, 0.1);
NormalizedIso = result.IsotopicProfile.CloneIsotopicProfile();
if (indexOfCorrespondingObservedPeak >= 0)
{
IsotopicProfileUtilities.NormalizeIsotopicProfileToSpecificPeak(NormalizedIso, indexOfCorrespondingObservedPeak);
}
else
{
IsotopicProfileUtilities.NormalizeIsotopicProfile(NormalizedIso);
}
//insert zero intensity peaks into observed
// return;
if (result.Flags.Count > 0)
{
var flagstring = "";
foreach (var resultFlag in result.Flags)
{
flagstring += resultFlag.Description + ";";
}
result.ErrorDescription = flagstring.TrimEnd(';');
result.FailedResult = true;
if (flagstring.Contains("PeakToTheLeft"))
{
result.FailureType = Globals.TargetedResultFailureType.DeisotopingProblemDetected;
}
else
{
result.FailureType = Globals.TargetedResultFailureType.QuantifierFailure;
}
}
var resultPassesMinimalCriteria = (result.Score < MaximumFitScoreForFurtherProcessing && result.Flags.Count == 0);
if (resultPassesMinimalCriteria)
{
//------------------------------------- Get chromatogramCorrelation data -------------------------------
if (IsChromatogramCorrelationPerformed)
{
GetChromatogramCorrelationData(result);
}
NormalizedAdjustedIso = NormalizedIso.CloneIsotopicProfile();
//this is experimental!! it attempts to correct the problems caused by Orbitraps on isotopic profile intensities
//UpdateIsoIntensitiesUsingChromCorrData(result.ChromCorrelationData, NormalizedAdjustedIso);
//----------------- create ratio data -------------------------------------------------------
var ratioData = NormalizedAdjustedIso.CloneIsotopicProfile();
for (var i = 0; i < NormalizedIso.Peaklist.Count; i++)
{
if (i < theorUnlabelledIso.Peaklist.Count && theorUnlabelledIso.Peaklist[i].Height > MinimumRelativeIntensityForRatioCalc)
{
ratioData.Peaklist[i].Height = (NormalizedIso.Peaklist[i].Height / theorUnlabelledIso.Peaklist[i].Height - 1);
}
else
{
ratioData.Peaklist[i].Height = 0;
}
}
//trim off zeros from ratio data
for (var i = ratioData.Peaklist.Count - 1; i >= 0; i--)
{
if (ratioData.Peaklist[i].Height == 0)
{
ratioData.Peaklist.RemoveAt(i);
}
else
{
break;
}
}
var xvals = ratioData.Peaklist.Select((p, i) => new { peak = p, index = i }).Select(n => (double)n.index).ToList();
var yvals = ratioData.Peaklist.Select(p => (double)p.Height).ToList();
result.AreaUnderRatioCurve = yvals.Sum();
RatioVals.Xvalues = xvals.ToArray();
RatioVals.Yvalues = yvals.ToArray();
//NOTE: Sept 23, 2013 - the 'R is no longer used.
result.RSquaredValForRatioCurve = GetLinearRegressionData(result, xvals, yvals);
//------------- subtract unlabelled profile from normalized profile ----------------------
var subtractedIsoData = NormalizedAdjustedIso.CloneIsotopicProfile();
for (var i = 0; i < subtractedIsoData.Peaklist.Count; i++)
{
float intensityTheorPeak = 0;
if (i < theorUnlabelledIso.Peaklist.Count)
{
intensityTheorPeak = theorUnlabelledIso.Peaklist[i].Height;
}
var subtractedIntensity = subtractedIsoData.Peaklist[i].Height - intensityTheorPeak;
if (subtractedIntensity < 0)
{
subtractedIntensity = 0;
}
subtractedIsoData.Peaklist[i].Height = subtractedIntensity;
}
result.AreaUnderDifferenceCurve = subtractedIsoData.Peaklist.Select(p => p.Height).Sum();
//----------- get data for the subtracted, labeled isotopic profile------------------------
var peaksForLabeledIsoQuant = new List <Peak>(subtractedIsoData.Peaklist.Where(p => p.Height > 0));
// var isoFromPartialLabelingQuantifier = _partialLabelingQuantifier.FindBestLabeledProfile(result.Target, peaksForLabeledIsoQuant);
//FitScoreData = _partialLabelingQuantifier.FitScoreData;
//result.FitScoreLabeledProfile = isoFromPartialLabelingQuantifier.IsotopicProfile == null ? 1.00d : isoFromPartialLabelingQuantifier.IsotopicProfile.Score;
//result.PercentCarbonsLabelled = isoFromPartialLabelingQuantifier.PercentLabeling;
//int numCarbons = result.Target.GetAtomCountForElement("C");
//result.NumCarbonsLabelled = result.PercentCarbonsLabelled * numCarbons / 100;
//StringBuilder sb = new StringBuilder();
//sb.Append(result.Target.ID + "\t" + result.Target.MZ.ToString("0.0000") + "\t" + result.Target.ChargeState +
// "-----------------------------\n");
//int counter = 0;
//foreach (var fsData in _partialLabelingQuantifier.FitScoreData)
//{
// sb.Append(fsData.Key + "\t" + fsData.Value.ToString("0.000") + "\n");
// counter++;
//}
//Console.WriteLine(sb.ToString());
//-------------- make calculations using inputs from chrom correlation data -------------------
HighQualitySubtractedProfile = GetIsoDataPassingChromCorrelation(result.ChromCorrelationData, subtractedIsoData);
var contiguousnessScore = GetContiguousnessScore(HighQualitySubtractedProfile);
result.ContiguousnessScore = contiguousnessScore;
//GORD ------------- note this section is a duplicate of the above.... choose one or the other ------------------------
//Feb 26, 2013... I processed the results and these look really good. ROC curve nice
peaksForLabeledIsoQuant = new List <Peak>(HighQualitySubtractedProfile.Peaklist.Where(p => p.Height > 0));
//calculate labeled fit score
var isoFromPartialLabelingQuantifier = _partialLabelingQuantifier.FindBestLabeledProfile(result.Target, peaksForLabeledIsoQuant);
FitScoreData = _partialLabelingQuantifier.FitScoreData;
result.FitScoreLabeledProfile = isoFromPartialLabelingQuantifier.IsotopicProfile == null ? 1.00d : isoFromPartialLabelingQuantifier.IsotopicProfile.Score;
result.PercentCarbonsLabelled = isoFromPartialLabelingQuantifier.PercentLabeling;
var numCarbons = result.Target.GetAtomCountForElement("C");
result.NumCarbonsLabelled = result.PercentCarbonsLabelled * numCarbons / 100;
//end of section --------------------------------------------------------------------------
//-------------- calculate Label Distribution ------------------------------------------------
// see Chik et al: http://pubs.acs.org/doi/abs/10.1021/ac050988l ; Also see Sipper paper.
double[] numLabelVals;
double[] labelDistributionVals;
var theorIntensityVals = theorUnlabelledIso.Peaklist.Select(p => (double)p.Height).ToList();
if (isoFromPartialLabelingQuantifier.IsotopicProfile != null)
{
var normalizedCorrectedIntensityVals = NormalizedAdjustedIso.Peaklist.Select(p => (double)p.Height).ToList();
var numRightPads = 3;
_labelingDistributionCalculator.CalculateLabelingDistribution(theorIntensityVals, normalizedCorrectedIntensityVals,
LabeldistCalcIntensityThreshold,
LabeldistCalcIntensityThreshold,
out numLabelVals,
out labelDistributionVals, true, true, 0, numRightPads, 0, 0);
//negative distribution values are zeroed out. And, then the remaining values are adjusted such that they add up to 1.
result.LabelDistributionVals = AdjustLabelDistributionVals(labelDistributionVals);
double distFractionUnlabelled, distFractionLabelled, distAverageLabelsIncorporated;
_labelingDistributionCalculator.OutputLabelingInfo(result.LabelDistributionVals, out distFractionUnlabelled,
out distFractionLabelled,
out distAverageLabelsIncorporated);
result.PercentPeptideLabelled = distFractionLabelled * 100;
}
else
{
result.PercentCarbonsLabelled = 0;
}
var highQualityRatioProfileData = GetIsoDataPassingChromCorrelation(result.ChromCorrelationData, ratioData);
if (highQualityRatioProfileData.Peaklist != null && highQualityRatioProfileData.Peaklist.Count > 0)
{
result.AreaUnderDifferenceCurve = HighQualitySubtractedProfile.Peaklist.Select(p => p.Height).Sum();
}
else
{
result.AreaUnderDifferenceCurve = 0;
}
if (HighQualitySubtractedProfile.Peaklist != null && HighQualitySubtractedProfile.Peaklist.Count > 0)
{
result.AreaUnderRatioCurveRevised = highQualityRatioProfileData.Peaklist.Select(p => p.Height).Sum();
}
else
{
result.AreaUnderRatioCurveRevised = 0;
}
result.NumHighQualityProfilePeaks = highQualityRatioProfileData.Peaklist != null
? highQualityRatioProfileData.Peaklist.Count
: 0;
}
else
{
result.FailedResult = true;
if (result.FailureType == Globals.TargetedResultFailureType.None)
{
result.FailureType = Globals.TargetedResultFailureType.MinimalCriteriaNotMet;
}
}
}