public static List <SpectrumData> ReadKnownMSMSSpectraList(string knownMSMSPath) // read all known compands MSMS data to a list
{
List <SpectrumData> allSpectrumList = new List <SpectrumData>();
List <string> txtList = ReadFileToList(knownMSMSPath);
for (int i = 0; i < txtList.Count - 1; i++)
{
if (txtList[i].Contains("eV:")) // indecating sampleName line
{
string[] tittleLine = txtList[i].Split(',');
SpectrumData standardSpectrum = new SpectrumData();
List <PeakData> spectrumPeakList = new List <PeakData>();
for (int j = i + 1; j < txtList.Count; j++)
{
string[] line = txtList[j].Split(default(char[]), StringSplitOptions.RemoveEmptyEntries);
if (line.GetLength(0) > 1)
{
PeakData peak = new PeakData();
peak.mz = double.Parse(line[0]);
peak.intensity = double.Parse(line[1]);
peak.sampleName = txtList[i].Replace(",", "_");
peak.rt = double.Parse(tittleLine[tittleLine.GetLength(0) - 2].Split(default(char[]), StringSplitOptions.RemoveEmptyEntries)[0]) * 60;
spectrumPeakList.Add(peak);
}
else
{
break;
}
}
standardSpectrum.peakList = spectrumPeakList;
allSpectrumList.Add(standardSpectrum);
}
}
return(allSpectrumList);
}
private static SpectrumData AverageIntensityofMZline(SpectrumData allPeakDataInEachGroup) // this method is to calculate average intensity for each mz line with the one 3 times higher than noise in one of the 20-50ev spectrum(calculated by pvals). to create a list containing each mz as list[i] with position of this mz
{
SpectrumData averageSpectrum = new SpectrumData();
averageSpectrum.peakList = new List <PeakData>();
var mzLines = from peaks in allPeakDataInEachGroup.peakList
group peaks by peaks.mz;
foreach (var mzline in mzLines)
{
double sumIntensity = 0, averageIntensity = 0;
int n = 0;
List <string> peakPositionList = new List <string>();
StringBuilder peakPosition = new StringBuilder();
foreach (var peak in mzline)
{
if (peak.isNoise == false)
{
sumIntensity += peak.intensity;
peakPositionList.Add(peak.sampleName.Split('_')[0]);
n++;
}
}
PeakData averPeak = mzline.ToList <PeakData>()[0]; // choose the first peak in each mz line
averPeak.isNoise = true;
if (n != 0)
{
averageIntensity = sumIntensity / n;
if (n >= minSampleNrForCor) // minimun samples for correlation analysis
{
averPeak.isNoise = false;
}
}
for (int i = 0; i < peakPositionList.Count; i++)
{
if (i == peakPositionList.Count - 1)
{
peakPosition.Append(peakPositionList[i]);
}
else
{
peakPosition.Append(peakPositionList[i] + "_");
}
}
averPeak.intensity = averageIntensity;
averPeak.sampleName = peakPosition.ToString(); //name formats as 53_196_115
averageSpectrum.peakList.Add(averPeak);
}
averageSpectrum.group = allPeakDataInEachGroup.group;
return(averageSpectrum);
}
private static bool IsMainPeak(PeakData iso, PeakData main)
{
bool reslt = false;
if (IsIsotope(iso.isotopes) && main.isotopes.Contains("[M]"))
{
int n = int.Parse(iso.isotopes.Split('+')[1].Replace("]", ""));
if (((iso.mz - main.mz) < 1.01 * n) && (iso.mz - main.mz) > 1 * n)
{
reslt = true;
}
}
return(reslt);
}
/**
* Merge multiple idMSMS (if mz difference less than 0.01, pick the higher intensity one)
* This function is used in two cases：1, merge 20-50ev idMSMSs， 2, merge idMSMSs in the same pcgroup with NDP>threshold
*/
private static SpectrumData MergeSpectraToOne(List <SpectrumData> spectraWithSamePcgrp)
{
SpectrumData maxSpectrum = new SpectrumData();
maxSpectrum.peakList = new List <PeakData>();
maxSpectrum.group = spectraWithSamePcgrp.OrderByDescending(x => double.Parse(x.group.Split('_')[0])).ToList()[0].group; //choose the highest mz as precursor(if this is 20-50ev, group is the same, then no influnce)
List <PeakData> allPeaksInOneGroup = new List <PeakData>();
for (int i = 0; i < spectraWithSamePcgrp.Count; i++)
{
allPeaksInOneGroup.AddRange(spectraWithSamePcgrp[i].peakList);
}
var allpeaks = from peaks in allPeaksInOneGroup
group peaks by peaks.mz;
foreach (var peaks in allpeaks)
{
PeakData maxPeak = peaks.OrderByDescending(x => x.intensity).ToList()[0]; //choose the highest intensity peak in one mz
maxSpectrum.peakList.Add(maxPeak);
}
for (int m = 0; m < maxSpectrum.peakList.Count; m++) //remove redundancy(two mz <0.01)
{
for (int n = m + 1; n < maxSpectrum.peakList.Count; n++)
{
if (Cal_Cosine_Product.IsEqualMZ(maxSpectrum.peakList[m].mz, maxSpectrum.peakList[n].mz))
{
if (maxSpectrum.peakList[m].intensity >= maxSpectrum.peakList[n].intensity)
{
maxSpectrum.peakList.RemoveAt(n);
n = n - 1;
}
else
{
maxSpectrum.peakList[m] = maxSpectrum.peakList[n];
n = n - 1;
}
}
}
}
for (int i = 0; i < maxSpectrum.peakList.Count; i++) //assign max precursor to all peaks
{
maxSpectrum.peakList[i].precursor = maxSpectrum.group;
}
maxSpectrum.peakList = maxSpectrum.peakList.OrderBy(x => x.mz).ToList();
return(maxSpectrum);
}
private static PeakData AverageIntensityofMZline(List <PeakData> mzPeakList) // this method is to calculate average intensity for each mz line with the one 3 times higher than noise in one of the 20-50ev spectrum(calculated by pvals). to create a list containing each mz as list[i] with position of this mz
{
PeakData averPeak = new PeakData();
double sumIntensity = 0, averageIntensity = 0;
int n = 0;
List <string> peakPositionList = new List <string>();
StringBuilder peakPosition = new StringBuilder();
for (int i = 0; i < mzPeakList.Count; i++)
{
if (mzPeakList[i].isNoise == false)
{
sumIntensity += mzPeakList[i].intensity;
peakPositionList.Add(mzPeakList[i].sampleName.Split('_')[0]);
n++;
}
}
averPeak = mzPeakList[0]; // choose the first peak in each mz line
averPeak.isNoise = true;
if (n != 0)
{
averageIntensity = sumIntensity / n;
if (n >= minSampleNrForCor) // minimun samples for correlation analysis
{
averPeak.isNoise = false;
}
}
for (int i = 0; i < peakPositionList.Count; i++)
{
if (i == peakPositionList.Count - 1)
{
peakPosition.Append(peakPositionList[i]);
}
else
{
peakPosition.Append(peakPositionList[i] + "_");
}
}
averPeak.intensity = averageIntensity;
averPeak.sampleName = peakPosition.ToString(); //name formats as 53_196_115
return(averPeak);
}
private static SpectrumData CalMaxAverageIntensity(SpectrumData allRelativeIntensity) // calculate maximum average intensity
{
SpectrumData maxAverageSpectrum = new SpectrumData();
maxAverageSpectrum.peakList = new List <PeakData>();
List <PeakData> spectrumList = new List <PeakData>();
var mz = from peaks in allRelativeIntensity.peakList
group peaks by peaks.mz;
foreach (var peaks in mz)
{
PeakData maxPeak = new PeakData();
List <PeakData> ev20 = new List <PeakData>(); //store 20ev samples
List <PeakData> ev30 = new List <PeakData>();
List <PeakData> ev40 = new List <PeakData>();
List <PeakData> ev50 = new List <PeakData>();
List <PeakData> ms = new List <PeakData>();
foreach (var peak in peaks)
{
if (peak.sampleName.Contains("20eV"))
{
ev20.Add(peak);
}
if (peak.sampleName.Contains("30eV"))
{
ev30.Add(peak);
}
if (peak.sampleName.Contains("40eV"))
{
ev40.Add(peak);
}
if (peak.sampleName.Contains("50eV"))
{
ev50.Add(peak);
}
if (peak.sampleName.Contains(".Ms."))
{
ms.Add(peak);
}
}
List <PeakData> mzAverageList = new List <PeakData>(); // store average intensity
if (ev20.Count != 0)
{
ev20[0].intensity = ev20.Average(x => x.intensity);
mzAverageList.Add(ev20[0]);
}
if (ev30.Count != 0)
{
ev30[0].intensity = ev30.Average(x => x.intensity);
mzAverageList.Add(ev30[0]);
}
if (ev40.Count != 0)
{
ev40[0].intensity = ev40.Average(x => x.intensity);
mzAverageList.Add(ev40[0]);
}
if (ev50.Count != 0)
{
ev50[0].intensity = ev50.Average(x => x.intensity);
mzAverageList.Add(ev50[0]);
}
if (ms.Count != 0)
{
ms[0].intensity = ms.Average(x => x.intensity);
mzAverageList.Add(ms[0]);
}
maxPeak = mzAverageList[0];
maxPeak.intensity = mzAverageList.Max(x => x.intensity);
maxPeak.sampleName = "MergedSpectrum" + maxPeak.pcgroup.ToString();
spectrumList.Add(maxPeak);
}
maxAverageSpectrum.peakList = spectrumList;
maxAverageSpectrum.group = spectrumList[0].precursor;
return(maxAverageSpectrum);
}
/**
* After using this function，list contains a <PeakData> List， each index in the list coresponds to one mz （value averaged）
*/
public static List <SpectrumData> ProcessPeaktableToSpectralist(string csvFilePath) //read result CSV file generated by XCMS and CAMERA to a list of spectra
{
List <SpectrumData> allSpectrumList = new List <SpectrumData>();
List <PeakData> allPeakList = new List <PeakData>();
List <string> CSVList = new List <string>();
CSVList = FileProcess.ReadFileToList(csvFilePath);
//CSVList = BigFileReader.GetFileContent(csvFilePath);
string[] titleLine = null;
if (CSVList.Count != 0) // extract tittle row
{
titleLine = CSVList[0].Split(',');
}
else
{
Console.WriteLine("CSVFile is empty!");
}
int diffrtIndex = 0, pvalIndex = 0, featureIndex = 0, mzIndex = 4, rtIndex = 5,
isotopesIndex = titleLine.GetLength(0) - 4, adductIndex = titleLine.GetLength(0) - 3, pcgroupIndex = titleLine.GetLength(0) - 2, precursorIndex = titleLine.GetLength(0) - 1;
//store the index of interested tittles, allocate initial default values.
for (int j = 0; j < titleLine.GetLength(0); j++)
{
titleLine[j] = titleLine[j].Replace("\"", "");
switch (titleLine[j])
{
case "mz":
mzIndex = j;
break;
case "rt":
rtIndex = j;
break;
case "isotopes":
isotopesIndex = j;
break;
case "adduct":
adductIndex = j;
break;
case "pcgroup":
pcgroupIndex = j;
break;
case "pval":
pvalIndex = j;
break;
case "feature":
featureIndex = j;
break;
case "diffrt":
diffrtIndex = j;
break;
case "precursor":
precursorIndex = j;
break;
default:
break;
}
}
for (int i = 1; i < CSVList.Count; i++) // extract data rows, each time extract 1 row and take title row as a pair.
{
string[] line = CSVList[i].Replace("\"", "").Split(',');
List <PeakData> mzPeaks = new List <PeakData>();
for (int j = 0; j < line.GetLength(0); j++)
{
if (titleLine[j].Contains("_") && titleLine[j].Length > 1) // all the sample names contain "_" char
{
PeakData peak = new PeakData();
peak.mz = double.Parse(line[mzIndex]);
peak.rt = double.Parse(line[rtIndex]);
peak.isotopes = line[isotopesIndex];
peak.adduct = line[adductIndex];
peak.pcgroup = int.Parse(line[pcgroupIndex]);
peak.intensity = double.Parse(line[j]);
peak.sampleName = titleLine[j];
if (diffrtIndex != 0) //MS doesnt contain pval, feature, diffrt
{
peak.pval = double.Parse(line[pvalIndex]);
peak.feature = double.Parse(line[featureIndex]);
peak.diffrt = double.Parse(line[diffrtIndex]);
}
peak.precursor = line[precursorIndex].Replace(" ", "");
if (double.Parse(line[j]) >= weight * bg) //intensity which is 3 times higher than the average noise will be deemed as real peak!
{
peak.isNoise = false;
}
else
{
peak.isNoise = true;
}
mzPeaks.Add(peak);
}
}
PeakData averPeak = AverageIntensityofMZline(mzPeaks);
allPeakList.Add(averPeak);// after AverageIntensityofMZline, only one peakdata left
}
var groupResult = from p in allPeakList
group p by p.precursor;
foreach (var groups in groupResult)
{
List <PeakData> peakList = new List <PeakData>();
peakList = groups.ToList <PeakData>();
SpectrumData spec = new SpectrumData();
spec.peakList = peakList; //spec contains all original peak data
spec.group = groups.Key.Replace(" ", "");
if (peakList.Count != 0) //0 means no peaks, 1means single peak group, should be removed
{
allSpectrumList.Add(spec); //here merged all samples and CIDs into allSpectrumList
}
else
{
continue;
}
}
allSpectrumList = allSpectrumList.OrderBy(x => x.group).ToList();
return(allSpectrumList);
}
public static double rtWindow = 9; // rt window to precursor
//
public static List <SpectrumData> MergeIdMSMSOfDiffCIDs(string rootDataPath) // using idMSMS precursor-product method by calculated pvals, this method will merge 20-50ev data by selecting highest intensity(the peaks are different among CIDs)
{
List <SpectrumData> mergedSpecList = new List <SpectrumData>();
List <SpectrumData> mergedCIDList = new List <SpectrumData>();
List <SpectrumData> AllCIDList = new List <SpectrumData>(); // all CID
List <SpectrumData> CID20evList = ProcessPeaktableToSpectralist(rootDataPath.Replace("MS", "20ev"));
AllCIDList.AddRange(CID20evList.Where(x => double.Parse(x.group.Split('_')[1]) > 50).ToList()); //remove rt<50
CID20evList = null;
GC.Collect();
Console.WriteLine("20ev done!");
List <SpectrumData> CID30evList = ProcessPeaktableToSpectralist(rootDataPath.Replace("MS", "30ev"));
AllCIDList.AddRange(CID30evList.Where(x => double.Parse(x.group.Split('_')[1]) > 50).ToList());
CID30evList = null;
GC.Collect();
Console.WriteLine("30ev done!");
List <SpectrumData> CID40evList = ProcessPeaktableToSpectralist(rootDataPath.Replace("MS", "40ev"));
AllCIDList.AddRange(CID40evList.Where(x => double.Parse(x.group.Split('_')[1]) > 50).ToList());
CID40evList = null;
GC.Collect();
Console.WriteLine("40ev done!");
List <SpectrumData> CID50evList = ProcessPeaktableToSpectralist(rootDataPath.Replace("MS", "50ev"));
AllCIDList.AddRange(CID50evList.Where(x => double.Parse(x.group.Split('_')[1]) > 50).ToList());
CID50evList = null;
GC.Collect();
Console.WriteLine("50ev done!");
List <SpectrumData> MSList = ProcessPeaktableToSpectralist(rootDataPath); // Note: pcgroup is different among different CIDs
List <PeakData> allMSPeakList = new List <PeakData>();
for (int i = 0; i < MSList.Count; i++)
{
allMSPeakList.AddRange(MSList[i].peakList.Where(x => x.rt > 50).ToList()); //already averaged, meaning that each PeakData=one mz line
}
MSList = null;
GC.Collect();
var allSpec = from specs in AllCIDList
group specs by specs.peakList[0].precursor;
foreach (var specs in allSpec)
{
PeakData precursorPeak = allMSPeakList.Where(x => x.mz == double.Parse(specs.Key.Split('_')[0])).ToList()[0]; // find this precursor in MS
double precursorPeakRT = precursorPeak.rt;
/**
* Filter presursor: 1,precursor should pass the "noise"threshold; 2, isotope is removed from precursor list（but isotopes can be fragments; 3, precursors will meet the criteria that "diffrt"(retention time difference) and "feature"(mass difference) are zero
*/
List <PeakData> peaksofeachprecursor = new List <PeakData>();
for (int k = 0; k < specs.ToList().Count; k++)
{
peaksofeachprecursor.AddRange(specs.ToList()[k].peakList);
}
if (precursorPeak.isNoise == true || IsIsotope(precursorPeak.isotopes) || !IsSelfCor(peaksofeachprecursor)) // determine whether precursor peak contains minimum number of "minSampleNrForCor" or is isotope, if not, not include
{
continue;
}
/**
* Filter fragments: 1, fragments should not be "noise" peaks，2, fragment's "sample position" should be the subcollection of precursor's position，3, fragment's m/z less than precursor（The highest intensity of fragment is selected eventually，and then calculate its relativeintensity）
*/
List <string> precursorPeakPosition = precursorPeak.sampleName.Split('_').ToList();
double precursorMZ = precursorPeak.mz;
List <PeakData> allPeaksInOneGroup = new List <PeakData>();
List <PeakData> MList = new List <PeakData>();
foreach (var spec in specs)
{
allPeaksInOneGroup.AddRange(spec.peakList.Where(x => (x.isNoise == false && x.diffrt < rtWindow)).ToList());// determine whether fragment peak contains minimum number of "minSampleNrForCor", if not, not include
}
for (int i = 0; i < allPeaksInOneGroup.Count; i++)
{
List <string> fragmentPeakPosition = allPeaksInOneGroup[i].sampleName.Split('_').ToList();
double fragmentMZ = allPeaksInOneGroup[i].mz; //until here, the higher mz is not removed, so fragment mz can be higher than precursor mz here
if (precursorPeakPosition.Intersect(fragmentPeakPosition).Count() < minSampleNrForCor) // min samples for correlation
{
if (allPeaksInOneGroup[i].isotopes.Contains("[M]")) //after saving [M]+, check whether it is in the isotope list，if so, add this [M]+
{
MList.Add(allPeaksInOneGroup[i]);
}
allPeaksInOneGroup.RemoveAt(i);
i = i - 1;
}
}
for (int i = 0; i < MList.Count; i++)
{
for (int j = 0; j < allPeaksInOneGroup.Count; j++)
{
if (IsMainPeak(allPeaksInOneGroup[j], MList[i]))
{
allPeaksInOneGroup.Add(MList[i]);
break;
}
}
}
SpectrumData maxSpectrum = new SpectrumData();
maxSpectrum.peakList = allPeaksInOneGroup;
maxSpectrum.group = specs.Key;
List <SpectrumData> sd = new List <SpectrumData>();
sd.Add(maxSpectrum);
maxSpectrum = MergeSpectraToOne(sd); // merge 20-50ev
maxSpectrum.peakList = maxSpectrum.peakList.OrderBy(x => x.mz).ToList();
if (maxSpectrum.peakList.Count != 0)
{
mergedCIDList.Add(CalRelativeIntensity(maxSpectrum)); //calculate relative intensity
}
}
AllCIDList = null;
GC.Collect();
Console.WriteLine("Filter done!");
/**
* merging sub-idMS/MS function: merge those idMSMSs which are in the same pcgroup && ndp>thresholdNDP， then remove higher mz than precursor, calculate relative intensity， calculate neutral loss
*/
int n = 0;
var pcgrps = mergedCIDList.GroupBy(x => int.Parse(x.group.Split('_')[2])); //same pcgroup
foreach (var specs in pcgrps)
{
List <SpectrumData> specList = specs.ToList();
if (specList.Count > 1) //check if the pcgroup contains nore than one idMSMS
{
List <SpectrumData> similarList = new List <SpectrumData>();
specList = specList.OrderBy(x => double.Parse(x.group.Split('_')[0])).ToList();//ascending order mz, merge from bottom up
for (int i = 0; i < specList.Count - 1; i++)
{
PeakData precursorPeak = allMSPeakList.Where(x => x.mz == double.Parse(specList[i].group.Split('_')[0])).ToList()[0]; // find this precursor in MS
if (!Data_Operation.IsPrecursor(precursorPeak.adduct)) // if the mz is not precursor, merge upwards
{
for (int j = i + 1; j < specList.Count; j++)
{
double ndp = 0;
ndp = Cal_Cosine_Product.CalCosineProduct(specList[i], specList[j]);
if (ndp > ndpThreshold)
{
List <SpectrumData> candiList = new List <SpectrumData>();
candiList.Add(specList[i]);
candiList.Add(specList[j]);
SpectrumData MeSpec = MergeSpectraToOne(candiList);
specList[j] = MeSpec; //merge a and b (merged result is stored in b and delete a)
specList.RemoveAt(i);
i--;
n++;
break;
}
}
}
}
}
for (int i = 0; i < specList.Count; i++)
{
SpectrumData removeSpec = RemoveMZHigherThanPrecursor(specList[i]); //remove higher mz than precursor
if (removeSpec.peakList.Count != 0)
{
SpectrumData finalSpec = CalRelativeIntensity(removeSpec);// cal relative intensity
finalSpec.peakList = finalSpec.peakList.OrderByDescending(x => x.mz).ToList();
finalSpec.neutralLossList = Data_Operation.FindNeutralLoss(finalSpec);
mergedSpecList.Add(finalSpec);
}
}
}
Console.WriteLine("n:" + n);
Console.WriteLine("merge done!");
return(mergedSpecList);
}