private bool CompareSpectraBinData(
clsCorrelation dataComparer,
clsMSSpectrum fragSpectrum,
clsBinnedData binnedSpectrum)
{
var xData = new List <float>(fragSpectrum.IonCount);
var yData = new List <float>(fragSpectrum.IonCount);
// Make a copy of the data, excluding any Reporter Ion data
for (var index = 0; index < fragSpectrum.IonCount; index++)
{
if (!clsUtilities.CheckPointInMZIgnoreRange(fragSpectrum.IonsMZ[index],
mReporterIons.MZIntensityFilterIgnoreRangeStart,
mReporterIons.MZIntensityFilterIgnoreRangeEnd))
{
xData.Add((float)(fragSpectrum.IonsMZ[index]));
yData.Add((float)(fragSpectrum.IonsIntensity[index]));
}
}
binnedSpectrum.BinnedDataStartX = dataComparer.BinStartX;
binnedSpectrum.BinSize = dataComparer.BinSize;
// Note that the data in xData and yData should have already been filtered to discard data points below the noise threshold intensity
var success = dataComparer.BinData(xData, yData, binnedSpectrum.BinnedIntensities, binnedSpectrum.BinnedIntensitiesOffset);
return(success);
}
private float CompareSpectra(
clsMSSpectrum fragSpectrum1,
clsMSSpectrum fragSpectrum2,
clsBinningOptions binningOptions,
bool considerOffsetBinnedData = true)
{
// Compares the two spectra and returns a similarity score (ranging from 0 to 1)
// Perfect match is 1; no similarity is 0
// Note that both the standard binned data and the offset binned data are compared
// If considerOffsetBinnedData = True, then the larger of the two scores is returned
// similarity scores is returned
//
// If an error, returns -1
var binnedSpectrum1 = new clsBinnedData();
var binnedSpectrum2 = new clsBinnedData();
try
{
var dataComparer = new clsCorrelation(binningOptions);
RegisterEvents(dataComparer);
const clsCorrelation.cmCorrelationMethodConstants eCorrelationMethod = clsCorrelation.cmCorrelationMethodConstants.Pearson;
// Bin the data in the first spectrum
var success = CompareSpectraBinData(dataComparer, fragSpectrum1, binnedSpectrum1);
if (!success)
{
return(-1);
}
// Bin the data in the second spectrum
success = CompareSpectraBinData(dataComparer, fragSpectrum2, binnedSpectrum2);
if (!success)
{
return(-1);
}
// Now compare the binned spectra
// Similarity will be 0 if either instance of BinnedIntensities has fewer than 5 data points
var similarity1 = dataComparer.Correlate(binnedSpectrum1.BinnedIntensities, binnedSpectrum2.BinnedIntensities, eCorrelationMethod);
if (!considerOffsetBinnedData)
{
return(similarity1);
}
var similarity2 = dataComparer.Correlate(binnedSpectrum1.BinnedIntensitiesOffset, binnedSpectrum2.BinnedIntensitiesOffset, eCorrelationMethod);
return(Math.Max(similarity1, similarity2));
}
catch (Exception ex)
{
ReportError("CompareSpectra: " + ex.Message, ex);
return(-1);
}
}
/// <summary>
/// When returnMax is false, determine the sum of the data within the search mass tolerance
/// When returnMaxis true, determine the maximum of the data within the search mass tolerance
/// </summary>
/// <param name="msSpectrum"></param>
/// <param name="searchMZ"></param>
/// <param name="searchToleranceHalfWidth"></param>
/// <param name="ionMatchCount"></param>
/// <param name="closestMZ"></param>
/// <param name="returnMax"></param>
/// <returns>The sum or maximum of the matching data; 0 if no matches</returns>
/// <remarks>
/// Note that this function performs a recursive search of msSpectrum.IonsMZ
/// It is therefore very efficient regardless of the number of data points in the spectrum
/// For sparse spectra, you can alternatively use FindMaxValueInMZRange
/// </remarks>
public double AggregateIonsInRange(
clsMSSpectrum msSpectrum,
double searchMZ,
double searchToleranceHalfWidth,
out int ionMatchCount,
out double closestMZ,
bool returnMax)
{
ionMatchCount = 0;
closestMZ = 0;
double ionSumOrMax = 0;
try
{
var smallestDifference = double.MaxValue;
if (msSpectrum.IonsMZ != null && msSpectrum.IonCount > 0)
{
if (SumIonsFindValueInRange(msSpectrum.IonsMZ, searchMZ, searchToleranceHalfWidth, out var indexFirst, out var indexLast))
{
for (var ionIndex = indexFirst; ionIndex <= indexLast; ionIndex++)
{
if (returnMax)
{
// Return max
if (msSpectrum.IonsIntensity[ionIndex] > ionSumOrMax)
{
ionSumOrMax = msSpectrum.IonsIntensity[ionIndex];
}
}
else
{
// Return sum
ionSumOrMax += msSpectrum.IonsIntensity[ionIndex];
}
var testDifference = Math.Abs(msSpectrum.IonsMZ[ionIndex] - searchMZ);
if (testDifference < smallestDifference)
{
smallestDifference = testDifference;
closestMZ = msSpectrum.IonsMZ[ionIndex];
}
}
ionMatchCount = indexLast - indexFirst + 1;
}
}
}
catch (Exception)
{
ionMatchCount = 0;
}
return(ionSumOrMax);
}
public void ReplaceData(clsMSSpectrum spectrum, int scanNumberOverride)
{
ScanNumber = spectrum.ScanNumber;
if (ScanNumber != scanNumberOverride)
{
ScanNumber = scanNumberOverride;
}
IonsMZ.Clear();
IonsIntensity.Clear();
if (IonsMZ.Capacity / 2 > spectrum.IonsMZ.Count)
{
IonsMZ.Capacity = spectrum.IonsMZ.Count;
IonsIntensity.Capacity = spectrum.IonsIntensity.Count;
}
IonsMZ.AddRange(spectrum.IonsMZ);
IonsIntensity.AddRange(spectrum.IonsIntensity);
}
private void ComputeNoiseLevelForMassSpectrum(
clsScanInfo scanInfo,
clsMSSpectrum msSpectrum,
clsBaselineNoiseOptions noiseThresholdOptions)
{
const bool IGNORE_NON_POSITIVE_DATA = true;
scanInfo.BaselineNoiseStats = clsMASICPeakFinder.InitializeBaselineNoiseStats(0, noiseThresholdOptions.BaselineNoiseMode);
if (noiseThresholdOptions.BaselineNoiseMode == clsMASICPeakFinder.eNoiseThresholdModes.AbsoluteThreshold)
{
scanInfo.BaselineNoiseStats.NoiseLevel = noiseThresholdOptions.BaselineNoiseLevelAbsolute;
scanInfo.BaselineNoiseStats.PointsUsed = 1;
}
else if (msSpectrum.IonCount > 0)
{
mPeakFinder.ComputeTrimmedNoiseLevel(
msSpectrum.IonsIntensity, 0, msSpectrum.IonCount - 1,
noiseThresholdOptions, IGNORE_NON_POSITIVE_DATA,
out var newBaselineNoiseStats);
scanInfo.BaselineNoiseStats = newBaselineNoiseStats;
}
}
public clsMSSpectrum Copy(clsMSSpectrum sourceSpectrum)
{
var newSpectrum = new clsMSSpectrum(sourceSpectrum.ScanNumber, sourceSpectrum.IonsMZ, sourceSpectrum.IonsIntensity, sourceSpectrum.IonsMZ.Count);
return(newSpectrum);
}
public bool ProcessAndStoreSpectrum(
clsScanInfo scanInfo,
DataInput.clsDataImport dataImportUtilities,
clsSpectraCache spectraCache,
clsMSSpectrum msSpectrum,
clsBaselineNoiseOptions noiseThresholdOptions,
bool discardLowIntensityData,
bool compressData,
double msDataResolution,
bool keepRawSpectrum)
{
var lastKnownLocation = "Start";
try
{
// Determine the noise threshold intensity for this spectrum
// Stored in scanInfo.BaselineNoiseStats
lastKnownLocation = "Call ComputeNoiseLevelForMassSpectrum";
ComputeNoiseLevelForMassSpectrum(scanInfo, msSpectrum, noiseThresholdOptions);
if (!keepRawSpectrum)
{
return(true);
}
// Discard low intensity data, but not for MRM scans
if (discardLowIntensityData && scanInfo.MRMScanType == ThermoRawFileReader.MRMScanTypeConstants.NotMRM)
{
// Discard data below the noise level or below the minimum S/N level
// If we are searching for Reporter ions, then it is important to not discard any of the ions in the region of the reporter ion m/z values
lastKnownLocation = "Call DiscardDataBelowNoiseThreshold";
dataImportUtilities.DiscardDataBelowNoiseThreshold(msSpectrum,
scanInfo.BaselineNoiseStats.NoiseLevel,
mReporterIons.MZIntensityFilterIgnoreRangeStart,
mReporterIons.MZIntensityFilterIgnoreRangeEnd,
noiseThresholdOptions);
scanInfo.IonCount = msSpectrum.IonCount;
}
if (compressData)
{
lastKnownLocation = "Call CompressSpectraData";
// Again, if we are searching for Reporter ions, then it is important to not discard any of the ions in the region of the reporter ion m/z values
CompressSpectraData(msSpectrum, msDataResolution,
mReporterIons.MZIntensityFilterIgnoreRangeStart,
mReporterIons.MZIntensityFilterIgnoreRangeEnd);
}
if (msSpectrum.IonCount > MAX_ALLOWABLE_ION_COUNT)
{
// Do not keep more than 50,000 ions
lastKnownLocation = "Call DiscardDataToLimitIonCount";
mSpectraFoundExceedingMaxIonCount += 1;
// Display a message at the console the first 10 times we encounter spectra with over MAX_ALLOWABLE_ION_COUNT ions
// In addition, display a new message every time a new max value is encountered
if (mSpectraFoundExceedingMaxIonCount <= 10 || msSpectrum.IonCount > mMaxIonCountReported)
{
Console.WriteLine();
Console.WriteLine(
"Note: Scan " + scanInfo.ScanNumber + " has " + msSpectrum.IonCount + " ions; " +
"will only retain " + MAX_ALLOWABLE_ION_COUNT + " (trimmed " +
mSpectraFoundExceedingMaxIonCount.ToString() + " spectra)");
mMaxIonCountReported = msSpectrum.IonCount;
}
dataImportUtilities.DiscardDataToLimitIonCount(msSpectrum,
mReporterIons.MZIntensityFilterIgnoreRangeStart,
mReporterIons.MZIntensityFilterIgnoreRangeEnd,
MAX_ALLOWABLE_ION_COUNT);
scanInfo.IonCount = msSpectrum.IonCount;
}
lastKnownLocation = "Call AddSpectrumToPool";
var success = spectraCache.AddSpectrumToPool(msSpectrum, scanInfo.ScanNumber);
return(success);
}
catch (Exception ex)
{
ReportError("Error in ProcessAndStoreSpectrum (LastKnownLocation: " + lastKnownLocation + ")", ex, clsMASIC.eMasicErrorCodes.InputFileDataReadError);
return(false);
}
}
private void CompressSpectraData(
clsMSSpectrum msSpectrum,
double msDataResolution,
double mzIgnoreRangeStart,
double mzIgnoreRangeEnd)
{
// First, look for blocks of data points that consecutively have an intensity value of 0
// For each block of data found, reduce the data to only retain the first data point and last data point in the block
//
// Next, look for data points in msSpectrum that are within msDataResolution units of one another (m/z units)
// If found, combine into just one data point, keeping the largest intensity and the m/z value corresponding to the largest intensity
if (msSpectrum.IonCount <= 1)
{
return;
}
// Look for blocks of data points that all have an intensity value of 0
var targetIndex = 0;
var index = 0;
while (index < msSpectrum.IonCount)
{
if (msSpectrum.IonsIntensity[index] < float.Epsilon)
{
var countCombined = 0;
for (var comparisonIndex = index + 1; comparisonIndex < msSpectrum.IonCount; comparisonIndex++)
{
if (msSpectrum.IonsIntensity[comparisonIndex] < float.Epsilon)
{
countCombined += 1;
}
else
{
break;
}
}
if (countCombined > 1)
{
// Only keep the first and last data point in the block
msSpectrum.IonsMZ[targetIndex] = msSpectrum.IonsMZ[index];
msSpectrum.IonsIntensity[targetIndex] = msSpectrum.IonsIntensity[index];
targetIndex += 1;
msSpectrum.IonsMZ[targetIndex] = msSpectrum.IonsMZ[index + countCombined];
msSpectrum.IonsIntensity[targetIndex] = msSpectrum.IonsIntensity[index + countCombined];
index += countCombined;
}
// Keep this data point since a single zero
else if (targetIndex != index)
{
msSpectrum.IonsMZ[targetIndex] = msSpectrum.IonsMZ[index];
msSpectrum.IonsIntensity[targetIndex] = msSpectrum.IonsIntensity[index];
}
}
// Note: targetIndex will be the same as index until the first time that data is combined (countCombined > 0)
// After that, targetIndex will always be less than index and we will thus always need to copy data
else if (targetIndex != index)
{
msSpectrum.IonsMZ[targetIndex] = msSpectrum.IonsMZ[index];
msSpectrum.IonsIntensity[targetIndex] = msSpectrum.IonsIntensity[index];
}
index += 1;
targetIndex += 1;
}
// Update .IonCount with the new data count
msSpectrum.ShrinkArrays(targetIndex);
// Step through the data, consolidating data within msDataResolution
// Note that we're copying in place rather than making a new, duplicate array
// If the m/z value is between mzIgnoreRangeStart and mzIgnoreRangeEnd, then we will not compress the data
targetIndex = 0;
index = 0;
while (index < msSpectrum.IonCount)
{
var countCombined = 0;
var bestMz = msSpectrum.IonsMZ[index];
// Only combine data if the first data point has a positive intensity value
if (msSpectrum.IonsIntensity[index] > 0)
{
var pointInIgnoreRange = clsUtilities.CheckPointInMZIgnoreRange(msSpectrum.IonsMZ[index], mzIgnoreRangeStart, mzIgnoreRangeEnd);
if (!pointInIgnoreRange)
{
for (var comparisonIndex = index + 1; comparisonIndex < msSpectrum.IonCount; comparisonIndex++)
{
if (clsUtilities.CheckPointInMZIgnoreRange(msSpectrum.IonsMZ[comparisonIndex], mzIgnoreRangeStart, mzIgnoreRangeEnd))
{
// Reached the ignore range; do not allow to be combined with the current data point
break;
}
if (msSpectrum.IonsMZ[comparisonIndex] - msSpectrum.IonsMZ[index] < msDataResolution)
{
if (msSpectrum.IonsIntensity[comparisonIndex] > msSpectrum.IonsIntensity[index])
{
msSpectrum.IonsIntensity[index] = msSpectrum.IonsIntensity[comparisonIndex];
bestMz = msSpectrum.IonsMZ[comparisonIndex];
}
countCombined += 1;
}
else
{
break;
}
}
}
}
// Note: targetIndex will be the same as index until the first time that data is combined (countCombined > 0)
// After that, targetIndex will always be less than index and we will thus always need to copy data
if (targetIndex != index || countCombined > 0)
{
msSpectrum.IonsMZ[targetIndex] = bestMz;
msSpectrum.IonsIntensity[targetIndex] = msSpectrum.IonsIntensity[index];
index += countCombined;
}
index += 1;
targetIndex += 1;
}
// Update .IonCount with the new data count
msSpectrum.ShrinkArrays(targetIndex);
}
