public ProteinSpectrumMatchSet[][] GroupAcrossRuns(List<ProteinSpectrumMatchSet>[] prsmGroup, INodeComparer<ProteinSpectrumMatchSet> prsmGroupComparer)
{
var nDataset = prsmGroup.Length;
var prsmSet = new NodeSet<ProteinSpectrumMatchSet>() { };
for (var i = 0; i < nDataset; i++)
{
var groupedPrsms = prsmGroup[i];
if (groupedPrsms == null) continue;
prsmSet.AddRange(groupedPrsms);
}
var alignedPrsms = prsmSet.ConnnectedComponents(prsmGroupComparer);
var alignedResult = new ProteinSpectrumMatchSet[alignedPrsms.Count][];
for (var i = 0; i < alignedResult.Length; i++) alignedResult[i] = new ProteinSpectrumMatchSet[nDataset];
for(var i = 0; i < alignedPrsms.Count; i++)
{
foreach (var set in alignedPrsms[i])
{
if (alignedResult[i][set.DataId] != null)
{
alignedResult[i][set.DataId].Merge(set);
//Console.WriteLine("[{4}] {0}-{1}...{2}-{3}", set.MinScanNum, set.MaxScanNum, alignedResult[i][set.DataId].MinScanNum, alignedResult[i][set.DataId].MaxScanNum, set.DataId);
}
else
{
alignedResult[i][set.DataId] = set;
}
}
}
return alignedResult;
}
public LcMsFeature(double repMass, int repCharge, double repMz, int repScanNum, double abundance,
int minCharge, int maxCharge, int minScan, int maxScan,
double minElution, double maxElution, double minNet = 0, double maxNet = 0)
{
Abundance = abundance;
RepresentativeMass = repMass;
RepresentativeCharge = repCharge;
RepresentativeMz = repMz;
RepresentativeScanNum = repScanNum;
MinCharge = (minCharge > 0) ? minCharge : repCharge;
MaxCharge = (maxCharge > 0) ? maxCharge : repCharge;
MinScanNum = (minScan > 0) ? minScan : repScanNum;
MaxScanNum = (maxScan > 0) ? maxScan : repScanNum;
MinElutionTime = minElution;
MaxElutionTime = maxElution;
MaxNet = maxNet;
MinNet = minNet;
ProteinSpectrumMatches = new ProteinSpectrumMatchSet(0);
}
public LcMsFeature(double repMass, int repCharge, double repMz, int repScanNum, double abundance,
int minCharge, int maxCharge, int minScan, int maxScan,
double minElution, double maxElution, double minNet = 0, double maxNet = 0)
{
Abundance = abundance;
RepresentativeMass = repMass;
RepresentativeCharge = repCharge;
RepresentativeMz = repMz;
RepresentativeScanNum = repScanNum;
MinCharge = (minCharge > 0) ? minCharge : repCharge;
MaxCharge = (maxCharge > 0) ? maxCharge : repCharge;
MinScanNum = (minScan > 0) ? minScan : repScanNum;
MaxScanNum = (maxScan > 0) ? maxScan : repScanNum;
MinElutionTime = minElution;
MaxElutionTime = maxElution;
MaxNet = maxNet;
MinNet = minNet;
ProteinSpectrumMatches = new ProteinSpectrumMatchSet(0);
}
public void TestQuantifyIdedProteoforms()
{
var methodName = MethodBase.GetCurrentMethod().Name;
Utils.ShowStarting(methodName);
const string rawFolder = @"\\proto-11\MSXML_Cache\PBF_Gen_1_193\2015_2";
const string promexOutFolder = @"D:\MassSpecFiles\UTEX\MSAlign";
const string msAlignResultFolder = @"D:\MassSpecFiles\UTEX\MSAlign";
if (!Directory.Exists(rawFolder))
{
Assert.Ignore(@"Skipping test {0} since folder not found: {1}", methodName, rawFolder);
}
var nDataset = 32;
var dataset = new string[nDataset];
for (var i = 0; i < nDataset; i++)
{
dataset[i] = string.Format("Syn_utex2973_Top_{0,2:D2}_TopDown_7May15_Bane_14-09-01RZ", i + 1);
//var rawFile = string.Format(@"{0}\{1}.pbf", rawFolder, dataset[i]);
}
var prsmReader = new ProteinSpectrumMatchReader(0.01);
var filesProcessed = 0;
var tolerance = new Tolerance(10);
for (var i = 0; i < dataset.Length; i++)
{
var rawFile = string.Format(@"{0}\{1}.pbf", rawFolder, dataset[i]);
if (!File.Exists(rawFile))
{
Console.WriteLine(@"Warning: Skipping file not found: {0}", rawFile);
continue;
}
var run = PbfLcMsRun.GetLcMsRun(rawFile);
var path = string.Format(@"{0}\{1}_MSAlign_ResultTable.txt", msAlignResultFolder, dataset[i]);
if (!File.Exists(path))
{
Console.WriteLine(@"Warning: Skipping file not found: {0}", path);
continue;
}
var prsmList = prsmReader.LoadIdentificationResult(path, ProteinSpectrumMatch.SearchTool.MsAlign);
filesProcessed++;
for (var j = 0; j < prsmList.Count; j++)
{
var match = prsmList[j];
match.ProteinId = match.ProteinName.Substring(match.ProteinName.IndexOf(ProteinNamePrefix) + ProteinNamePrefix.Length, 5);
}
// PrSM To Feature
var prsmToFeatureIdMap = new int[prsmList.Count];
for (var k = 0; k < prsmToFeatureIdMap.Length; k++)
{
prsmToFeatureIdMap[k] = -1;
}
// Feature To PrSM
var featureToPrsm = new List <ProteinSpectrumMatchSet>();
var featureFinder = new LcMsPeakMatrix(run, new LcMsFeatureLikelihood());
var featureList = new List <LcMsPeakCluster>();
var featureId = 0;
for (var j = 0; j < prsmList.Count; j++)
{
if (prsmToFeatureIdMap[j] >= 0)
{
continue;
}
var match = prsmList[j];
var minScanNum = match.ScanNum;
var maxScanNum = match.ScanNum;
var mass = match.Mass;
var charge = match.Charge;
var massTh = tolerance.GetToleranceAsMz(mass);
var id1 = match.ProteinId;
var feature = featureFinder.GetLcMsPeakCluster(mass, charge, minScanNum, maxScanNum);
var prsmSet = new ProteinSpectrumMatchSet(i)
{
match
};
if (feature == null)
{
feature = featureFinder.GetLcMsPeaksFromNoisePeaks(mass, charge, minScanNum, maxScanNum, charge, charge);
prsmToFeatureIdMap[j] = featureId;
}
else
{
prsmToFeatureIdMap[j] = featureId;
var etTol = Math.Max(run.GetElutionTime(run.MaxLcScan) * 0.005, feature.ElutionLength * 0.2);
for (var k = j + 1; k < prsmList.Count; k++)
{
var otherMatch = prsmList[k];
var id2 = otherMatch.ProteinId;
var et2 = run.GetElutionTime(otherMatch.ScanNum);
if (id1.Equals(id2) &&
feature.MinElutionTime - etTol < et2 && et2 < feature.MaxElutionTime - etTol &&
Math.Abs(otherMatch.Mass - mass) < massTh)
{
prsmToFeatureIdMap[k] = featureId;
prsmSet.Add(otherMatch);
}
}
}
featureId++;
feature.Flag = 1;
featureList.Add(feature);
featureToPrsm.Add(prsmSet);
}
// Overlap between features???
for (var j = 0; j < featureList.Count; j++)
{
var f1 = featureList[j];
if (f1.Flag < 1)
{
continue;
}
var prsm1 = featureToPrsm[j];
for (var k = j + 1; k < featureList.Count; k++)
{
var f2 = featureList[k];
if (f2.Flag < 1)
{
continue;
}
var prsm2 = featureToPrsm[k];
if (Math.Abs(f1.Mass - f2.Mass) > tolerance.GetToleranceAsMz(f1.Mass))
{
continue;
}
if (!f1.CoElutedByNet(f2, 0.005))
{
continue;
}
if (!prsm1.ShareProteinId(prsm2))
{
continue;
}
// let us merge!!
if (f1.ScanLength > f2.ScanLength)
{
prsm1.AddRange(prsm2);
prsm2.Clear();
f2.Flag = 0;
}
else
{
prsm2.AddRange(prsm1);
prsm1.Clear();
f1.Flag = 0;
}
}
}
// now output results!!
var ms1ftFilePath = string.Format(@"{0}\{1}.ms1ft", promexOutFolder, dataset[i]);
var writer = new StreamWriter(ms1ftFilePath);
writer.WriteLine(LcMsFeatureFinderLauncher.GetHeaderString());
for (var j = 0; j < featureList.Count; j++)
{
var f1 = featureList[j];
if (f1.Flag < 1)
{
continue;
}
var prsm1 = featureToPrsm[j];
var minScanNum = run.GetPrevScanNum(prsm1.MinScanNum, 1);
var maxScanNum = run.GetNextScanNum(prsm1.MaxScanNum, 1);
f1.ExpandScanRange(minScanNum, maxScanNum);
writer.Write("{0}\t", j + 1);
writer.WriteLine(LcMsFeatureFinderLauncher.GetString(f1));
}
writer.Close();
Console.WriteLine(ms1ftFilePath);
}
if (filesProcessed == 0)
{
Assert.Ignore("Skipped since data files not found");
}
}