public static void WriteNDPResultForSytoscape(List <SpectrumData> allSpecList, string outputPath) //output NDP for Sytoscape input
{
using (StreamWriter sw = new StreamWriter(outputPath)) {
for (int i = 0; i < allSpecList.Count - 1; i++)
{
for (int j = i + 1; j < allSpecList.Count; j++)
{
double ndp = Cal_Cosine_Product.CalCosineProduct(allSpecList[i], allSpecList[j]);
if (ndp >= 0.01)
{
List <double> commonNeutralLostList = new List <double>();
for (int m = 0; m < allSpecList[i].neutralLossList.Count; m++)
{
for (int n = 0; n < allSpecList[j].neutralLossList.Count; n++)
{
if (Cal_Cosine_Product.IsEqualMZ(allSpecList[i].neutralLossList[m], allSpecList[j].neutralLossList[n])) // judge whether 2 mz values <=0.01
{
commonNeutralLostList.Add(Math.Round(allSpecList[i].neutralLossList[m], 2));
}
}
}
commonNeutralLostList = commonNeutralLostList.Distinct().ToList();
commonNeutralLostList = commonNeutralLostList.OrderBy(x => x).ToList();
sw.Write(allSpecList[i].group.Replace(" ", "") + "," + ndp + "," + allSpecList[j].group.Replace(" ", "") + ",");
for (int k = 0; k < commonNeutralLostList.Count; k++)
{
sw.Write(commonNeutralLostList[k] + ",");
}
sw.WriteLine();
}
}
}
}
}
/**
* 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);
}
static void Main(string[] args)
{
Stopwatch watch = new Stopwatch();
watch.Start();
string rootPath = @"E:\Different tissues\idMSMS different tissue\MIX Results\";
/**
* Fill 0 values in XCMS peaktable
*/
//string fillpeaksPath = rootPath + "NormPeakArea.csv";
//string nonefillpeaksPath = rootPath + "data_minfrac0.1_sn3_nonfill.csv";
//string outPath = rootPath + "NormPeakArea_fill0.csv";
//Data_Operation.Fill0Peaks(fillpeaksPath, nonefillpeaksPath, outPath);
/**
* export precursors (change parapeters to export all/without isotope/only adducts)
*/
//string CSVFilePath = rootPath + "NormPeakArea.csv";
//List<SpectrumData> allSpectraList = IdMSMS_Deconvolution.ProcessPeaktableToSpectralist(CSVFilePath);
//Data_Operation.FindPrecursorMZ(allSpectraList, outPrecursorPath);
/**
* Assemble idMS/MS with multi-eV
*/
string CSVFilePath = rootPath + "AllDeconvolutedSpectra_MS_pvals_9s.csv";
string outputFilePath = rootPath + "idmsms_3_6_0.9_minCor_rmv50_50ev.csv";
string NLOutputFilePath = rootPath + "NL_motif.csv";
string NDPmatrixPath = rootPath + "NDPmatrix.csv";
List <SpectrumData> allMergedList = IdMSMS_Deconvolution.MergeIdMSMSOfDiffCIDs(CSVFilePath);
FileProcess.WriteSpectrumList(allMergedList, outputFilePath); //output idMSMS spectra
FileProcess.OutputNeutralLoss(allMergedList, NLOutputFilePath); //output NL_motif
string commonNLPath = rootPath + "Common NL table.csv";
string outNLHitPath = rootPath + "NLmatrix.csv";
FileProcess.OutputSelectedNLto360Spec(allMergedList, commonNLPath, outNLHitPath); //output binary NL hit table
//List<SpectrumData> allSpectraList = FileProcess.ReadSpectralCSVFile(CSVFilePath); // use if there is spectra list to read
//List<SpectrumData> allKnownSpectraList = FileProcess.ReadKnownMSMSSpectraList(knownMSMSPath); // use if there is known spectra list
List <SpectrumData> allKnownSpectraList = allMergedList;
List <double[]> NDPResultList = new List <double[]>();
string[] queryTitleLine = new string[allMergedList.Count]; // add QueryTitle line to result
for (int i = 0; i < allMergedList.Count; i++)
{
queryTitleLine[i] = allMergedList[i].group.ToString();
}
string[] libraryTitleLine = new string[allKnownSpectraList.Count]; // add Library title to result
for (int i = 0; i < allKnownSpectraList.Count; i++)
{
libraryTitleLine[i] = allKnownSpectraList[i].peakList[0].sampleName;
}
for (int i = 0; i < allMergedList.Count; i++)
{
double[] line = new double[allMergedList.Count]; //
for (int j = 0; j < allMergedList.Count; j++) //
{
line[j] = Cal_Cosine_Product.CalCosineProduct(allMergedList[i], allMergedList[j]); //
}
NDPResultList.Add(line);
}
FileProcess.WriteNDPListToFile(NDPResultList, queryTitleLine, queryTitleLine, NDPmatrixPath); // output NDPmatrix
watch.Stop();
Console.WriteLine("weight: " + IdMSMS_Deconvolution.weight + " minSamp:" + IdMSMS_Deconvolution.minSampleNrForCor + " ndpThreh:" + IdMSMS_Deconvolution.ndpThreshold);
Console.WriteLine("Deconvoluted spectra number: " + allMergedList.Count);
Console.WriteLine("calculation time: " + watch.Elapsed);
Console.WriteLine("OK");
Console.Read();
}
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);
}
