protected virtual void GetRawData(out List <double> vectMZs, out List <double> vectIntensities, int scan,
double min_mz, double max_mz, bool centroid)
{
if (max_mz <= min_mz)
{
throw new Exception("max_mz should be greater than min_mz");
}
vectMZs = new List <double>();
vectIntensities = new List <double>();
List <double> allMZs;
List <double> allIntensities;
GetRawData(out allMZs, out allIntensities, scan, centroid);
var numPts = allMZs.Count;
if (numPts <= 1)
{
Console.Error.WriteLine(scan);
throw new Exception("mz_vector empty in GetRawData in RawData.cpp");
}
var startIndex = PeakIndex.GetNearestBinary(allMZs, min_mz, 0, numPts - 1);
var stopIndex = PeakIndex.GetNearestBinary(allMZs, max_mz, 0, numPts - 1);
if ((stopIndex - startIndex) <= 1) //nothing in this m/z range
{
return;
}
vectMZs.AddRange(allMZs.GetRange(startIndex, stopIndex - startIndex));
vectIntensities.AddRange(allIntensities.GetRange(startIndex, stopIndex - startIndex));
}
/// <summary>
/// Gets the charge state (determined by AutoCorrelation algorithm) for a peak in some data.
/// </summary>
/// <param name="peak">is the peak whose charge we want to detect.</param>
/// <param name="peakData">is the PeakData object containing raw data, peaks, etc which are used in the process.</param>
/// <param name="debug"></param>
/// <returns>Returns the charge of the feature.</returns>
public static int GetChargeState(ThrashV1Peak peak, PeakData peakData, bool debug)
{
var minus = 0.1;
var plus = 1.1; // right direction to look
var startIndex = PeakIndex.GetNearest(peakData.MzList, peak.Mz - peak.FWHM - minus, peak.DataIndex);
var stopIndex = PeakIndex.GetNearest(peakData.MzList, peak.Mz + peak.FWHM + plus, peak.DataIndex);
var numPts = stopIndex - startIndex;
var numL = numPts;
if (numPts < 5)
{
return(-1);
}
if (numPts < 256)
{
numL = 10 * numPts;
}
// TODO: PattersonChargeStateCalculator does a lot of funny stuff around here.
// variable to help us perform spline interpolation.
// count is stopIndex - startIndex + 1 because we need to include the values at stopIndex as well
// NOTE: This output is different from what the DeconEngineV2 code outputs; there are notes in that code
// wondering if there was a bug in the previous implementation imported from VB, of starting at startIndex + 1.
// That code performed interpolation on the range from startIndex + 1 to stopIndex, inclusive, and minMz set to PeakData.MzList[startIndex + 1].
// This code can produce output that more closely matches the DeconEngineV2 output by using
// "startIndex, stopIndex - startIndex + 2" as the parameters to "GetRange()", and setting minMz to PeakData.MzList[startIndex + 1].
// Since using startIndex and stopIndex directly produces output that mostly differs on fit score by a small amount,
// we are changing this code to use them.
var interpolator = CubicSpline.InterpolateNaturalSorted(
peakData.MzList.GetRange(startIndex, stopIndex - startIndex + 1).ToArray(),
peakData.IntensityList.GetRange(startIndex, stopIndex - startIndex + 1).ToArray());
var minMz = peakData.MzList[startIndex];
var maxMz = peakData.MzList[stopIndex];
// List to store the interpolated intensities of the region on which we performed the cubic spline interpolation.
var iv = new List <double>(numL);
for (var i = 0; i < numL; i++)
{
var xVal = minMz + (maxMz - minMz) * i / numL;
var fVal = interpolator.Interpolate(xVal);
iv.Add(fVal);
}
if (debug)
{
Console.Error.WriteLine("mz,intensity");
for (var i = 0; i < numL; i++)
{
var xVal = minMz + (maxMz - minMz) * i / numL;
Console.Error.WriteLine(xVal + "," + iv[i]);
}
}
// List to store the auto correlation values at the points in the region.
var autoCorrelationScores = ACss(iv.ToArray());
if (debug)
{
Console.Error.WriteLine("AutoCorrelation values");
for (var i = 0; i < autoCorrelationScores.Count; i++)
{
var score = autoCorrelationScores[i];
Console.Error.WriteLine((maxMz - minMz) * i / numL + "," + score);
}
}
var minN = 0;
while (minN < numL - 1 && autoCorrelationScores[minN] > autoCorrelationScores[minN + 1])
{
minN++;
}
var success = HighestChargeStatePeak(minMz, maxMz, minN, autoCorrelationScores, MaxCharge, out var bestAcScore, out _);
if (!success)
{
return(-1); // Didn't find anything
}
// List to temporarily store charge list. These charges are calculated at peak values of auto correlation.
// Now go back through the CS peaks and make a list of all CS that are at least 10% of the highest
var charges = GenerateChargeStates(minMz, maxMz, minN, autoCorrelationScores, MaxCharge, bestAcScore);
// Get the final CS value to be returned
var returnChargeStateVal = -1;
// TODO: PattersonChargeStateCalculator really doesn't match the following code.
var fwhm = peak.FWHM; // Store a copy of the FWHM to avoid modifying the actual value
if (fwhm > 0.1)
{
fwhm = 0.1;
}
for (var i = 0; i < charges.Count; i++)
{
// no point retesting previous charge.
var tempChargeState = charges[i];
var skip = false;
for (var j = 0; j < i; j++)
{
if (charges[j] == tempChargeState)
{
skip = true;
break;
}
}
if (skip)
{
continue;
}
if (tempChargeState > 0)
{
var peakA = peak.Mz + 1.0 / tempChargeState;
var found = peakData.GetPeakFromAllOriginalIntensity(peakA - fwhm, peakA + fwhm, out var isoPeak);
if (found)
{
returnChargeStateVal = tempChargeState;
if (isoPeak.Mz * tempChargeState < 3000)
{
break;
}
// if the mass is greater than 3000, lets make sure that multiple isotopes exist.
peakA = peak.Mz - 1.03 / tempChargeState;
found = peakData.GetPeakFromAllOriginalIntensity(peakA - fwhm, peakA + fwhm, out isoPeak);
if (found)
{
return(tempChargeState);
}
}
else
{
peakA = peak.Mz - 1.0 / tempChargeState;
found = peakData.GetPeakFromAllOriginalIntensity(peakA - fwhm, peakA + fwhm, out isoPeak);
if (found && isoPeak.Mz * tempChargeState < 3000)
{
return(tempChargeState);
}
}
}
}
return(returnChargeStateVal);
}
public bool FindPeak(double minMz, double maxMz, out double mzValue, out double intensity, bool debug)
{
if (!_lastValueWasCached)
{
var found = IsotopeDistribution.FindPeak(minMz, maxMz, out mzValue, out intensity);
return(found);
}
mzValue = 0;
intensity = 0;
var numPts = TheoreticalDistMzs.Count;
var index = PeakIndex.GetNearestBinary(TheoreticalDistMzs, minMz, 0, numPts - 1);
if (index >= numPts)
{
return(false);
}
if (TheoreticalDistMzs[index] > minMz)
{
while (index > 0 && TheoreticalDistMzs[index] > minMz)
{
index--;
}
}
else
{
while (index < numPts && TheoreticalDistMzs[index] < minMz)
{
index++;
}
index--;
}
//find index of peak with maximum intensity
var maxIndex = -1;
for (; index < numPts; index++)
{
var mz = TheoreticalDistMzs[index];
if (mz > maxMz)
{
break;
}
if (mz > minMz)
{
if (TheoreticalDistIntensities[index] > intensity)
{
maxIndex = index;
intensity = TheoreticalDistIntensities[index];
}
}
}
if (maxIndex == -1)
{
return(false);
}
var x2 = TheoreticalDistMzs[maxIndex];
var x1 = x2 - 1.0 / IsotopeDistribution.PointsPerAmu;
var x3 = x2 + 1.0 / IsotopeDistribution.PointsPerAmu;
if (maxIndex > 0 && maxIndex < numPts - 1)
{
var y1 = TheoreticalDistIntensities[maxIndex - 1];
var y2 = TheoreticalDistIntensities[maxIndex];
var y3 = TheoreticalDistIntensities[maxIndex + 1];
// if the points are cached, these could be single sticks with surrounding
// points below background. To avoid that case, lets just check
// and see if the differences in the theoretical mz values is as
// expected or not.
if (TheoreticalDistMzs[maxIndex - 1] > x2 - 2.0 / IsotopeDistribution.PointsPerAmu &&
TheoreticalDistMzs[maxIndex + 1] < x2 + 2.0 / IsotopeDistribution.PointsPerAmu)
{
var d = (y2 - y1) * (x3 - x2); //[gord] slope? what is this?... Denominator... see below
d = d - (y3 - y2) * (x2 - x1); //
if (d.Equals(0))
{
mzValue = x2;
}
else
{
mzValue = (x1 + x2 - (y2 - y1) * (x3 - x2) * (x1 - x3) / d) / 2.0;
}
//[gord] what's this doing?? Looks like a mid-point calculation
}
else
{
mzValue = x2;
intensity = TheoreticalDistIntensities[maxIndex];
}
return(true);
}
mzValue = x2;
intensity = TheoreticalDistIntensities[maxIndex];
return(true);
}
/// <summary>
/// gets the intensity for a given mz.
/// </summary>
/// <remarks>
/// We look for the intensity at a given m/z value in the raw data List mzs
/// (the intensities are stored in the corresponding raw data intensity List intensities).
/// If the value does not exist, we interpolate the intensities of points before and after this m/z value.
/// </remarks>
/// <param name="mz">the m/z value for which we want to find the intensity.</param>
/// <param name="mzs">pointer to List with observed m/z values.</param>
/// <param name="intensities">pointer to List with observed intensity values.</param>
/// <returns>
/// returns the intensity of the peak which has the given m/z. If the exact peak is not present,
/// then we interpolate the intensity. If the m/z value is greater than the maximum mz or less,
/// then the minimum m/z, 0 is returned.
/// </returns>
public double GetPointIntensity(double mz, List <double> mzs, List <double> intensities)
{
if ((Mz1 < mz && Mz2 < mz) || (Mz1 > mz && Mz2 > mz))
{
var numpts = mzs.Count;
if (LastPointIndex >= numpts)
{
LastPointIndex = -1;
}
//since points are more likely to be searched in order.
if (LastPointIndex != -1 && Mz2 < mz)
{
if (LastPointIndex < numpts - 1 && mzs[LastPointIndex + 1] > mz)
{
LastPointIndex++;
}
else
{
LastPointIndex = PeakIndex.GetNearestBinary(mzs, mz, LastPointIndex,
numpts - 1);
}
}
else
{
LastPointIndex = PeakIndex.GetNearestBinary(mzs, mz, 0, numpts - 1);
}
if (LastPointIndex >= numpts)
{
return(0);
}
if (LastPointIndex < 0)
{
LastPointIndex = 0;
}
if (mzs[LastPointIndex] > mz)
{
while (LastPointIndex > 0 && mzs[LastPointIndex] > mz)
{
LastPointIndex--;
}
}
else
{
while (LastPointIndex < numpts && mzs[LastPointIndex] < mz)
{
LastPointIndex++;
}
LastPointIndex--;
}
if (LastPointIndex == numpts - 1)
{
LastPointIndex = numpts - 2;
}
Mz1 = mzs[LastPointIndex];
Mz2 = mzs[LastPointIndex + 1];
Intensity1 = intensities[LastPointIndex];
Intensity2 = intensities[LastPointIndex + 1];
}
if (Mz1.Equals(Mz2))
{
return(Intensity1);
}
return((mz - Mz1) / (Mz2 - Mz1) * (Intensity2 - Intensity1) + Intensity1);
}
