public void CollectTrainingSet()
{
var methodName = MethodBase.GetCurrentMethod().Name;
TestUtils.ShowStarting(methodName);
const string idFileFolder = @"D:\MassSpecFiles\training\IdResult";
const string outFileFolder = @"D:\MassSpecFiles\training\FilteredIdResult";
if (!Directory.Exists(idFileFolder))
{
Assert.Ignore(@"Skipping test {0} since folder not found: {1}", methodName, idFileFolder);
}
for (var d = 0; d < TrainSetFileLists.Length; d++)
{
var dataset = TrainSetFileLists[d];
var dataname = Path.GetFileNameWithoutExtension(dataset);
var idFile = string.Format(@"{0}\{1}_IcTda.tsv", idFileFolder, dataname);
var outFileName = string.Format(@"{0}\{1}.trainset.tsv", outFileFolder, Path.GetFileNameWithoutExtension(dataset));
if (File.Exists(outFileName))
{
continue;
}
Console.WriteLine(dataset);
if (!File.Exists(idFile))
{
idFile = string.Format(@"{0}\{1}_msgfdb_syn.txt", idFileFolder, dataname);
if (!File.Exists(idFile))
{
Console.WriteLine(@"Skipping file since not found: " + idFile);
continue;
}
}
Console.WriteLine(idFile);
var targetSets = LcMsFeatureTrain.CollectTrainSet(dataset, idFile);
Console.WriteLine(targetSets.Count);
var writer =
new StreamWriter(outFileName);
writer.WriteLine("MinScan\tMaxScan\tMinCharge\tMaxCharge\tMass\tSequence\tModifications");
foreach (var prsmSet in targetSets)
{
writer.Write(prsmSet.MinScanNum);
writer.Write("\t");
writer.Write(prsmSet.MaxScanNum);
writer.Write("\t");
writer.Write(prsmSet.MinCharge);
writer.Write("\t");
writer.Write(prsmSet.MaxCharge);
writer.Write("\t");
writer.Write(prsmSet.Mass);
writer.Write("\t");
writer.Write(prsmSet[0].Sequence);
writer.Write("\t");
writer.Write(prsmSet[0].Modifications);
//writer.Write("\t")
//writer.Write(string.Join(";", prsmSet.Select(prsm => prsm.ScanNum)));
writer.Write("\n");
}
writer.Close();
}
}
public void TestGenerateFrequencyData()
{
var methodName = MethodBase.GetCurrentMethod().Name;
Utils.ShowStarting(methodName);
const string idFileFolder = @"D:\MassSpecFiles\training\IdScoring\MSPF_trainset";
const string outFileFolder = @"D:\MassSpecFiles\training\IdScoring";
if (!Directory.Exists(idFileFolder))
{
Assert.Ignore(@"Skipping test {0} since folder not found: {1}", methodName, idFileFolder);
}
Modification.RegisterAndGetModification(Modification.Cysteinyl.Name, Modification.Cysteinyl.Composition);
Modification.RegisterAndGetModification(Modification.Phosphorylation.Name, Modification.Phosphorylation.Composition);
Modification.RegisterAndGetModification(Modification.Methylation.Name, Modification.Methylation.Composition);
Modification.RegisterAndGetModification(Modification.DiMethylation.Name, Modification.DiMethylation.Composition);
Modification.RegisterAndGetModification(Modification.TriMethylation.Name, Modification.TriMethylation.Composition);
Modification.RegisterAndGetModification("Trioxidation", new Composition(0, 0, 0, 3, 0));
// var aaSet = new AminoAcidSet(@"D:\MassSpecFiles\training\Mods.txt");
foreach (var dataset in TrainSetFileLists)
{
var dataname = Path.GetFileNameWithoutExtension(dataset);
var idFile = string.Format(@"{0}\{1}_IcTda.tsv", idFileFolder, dataname);
var decoyFile = string.Format(@"{0}\{1}_IcDecoy.tsv", idFileFolder, dataname);
// var targetFile = string.Format(@"{0}\{1}_IcTarget.tsv", idFileFolder, dataname);
if (!File.Exists(idFile))
{
continue;
}
var prsmReader = new ProteinSpectrumMatchReader(0.01);
var prsmList = prsmReader.LoadIdentificationResult(idFile);
var minScore = prsmList.Last().Score;
var decoyMatches = prsmReader.ReadMsPathFinderResult(decoyFile, int.MaxValue, 1, Math.Max(minScore - 5, 10));
var run = PbfLcMsRun.GetLcMsRun(dataset);
var spectrumMatchSet = LcMsFeatureTrain.CollectTrainSet(dataset, idFile);
Console.WriteLine(spectrumMatchSet.Count);
var writer = new StreamWriter(string.Format(@"{0}\{1}_target.tsv", outFileFolder, dataname));
foreach (var matches in spectrumMatchSet)
{
foreach (var match in matches)
{
var spec = run.GetSpectrum(match.ScanNum) as ProductSpectrum;
GetMatchStatistics(spec, match.GetSequence(), match.Charge, writer);
}
}
writer.Close();
writer = new StreamWriter(string.Format(@"{0}\{1}_decoy.tsv", outFileFolder, dataname));
foreach (var match in decoyMatches)
{
var sequence = match.GetSequence();
var spec = run.GetSpectrum(match.ScanNum) as ProductSpectrum;
GetMatchStatistics(spec, sequence, match.Charge, writer);
}
writer.Close();
}
}