public void TestParse()
{
var result = IdentificationSummary.Parse(@TestContext.CurrentContext.TestDirectory + "/../../../data/buildsummary/hppp_mascot.noredundant", "^REV", new TotalFalseDiscoveryRateCalculator());
Assert.AreEqual(381, result.FullSpectrumCount);
Assert.AreEqual(62, result.FullPeptideCount);
Assert.AreEqual(31, result.ProteinGroupCount);
Assert.AreEqual(16, result.Unique2ProteinGroupCount);
}
public static IdentificationSummary Parse(string proteinFile, string defaultDecoyPattern, IFalseDiscoveryRateCalculator defaultCalc)
{
IdentificationSummary result = new IdentificationSummary();
result.FileName = FileUtils.ChangeExtension(new FileInfo(proteinFile).Name, "");
Regex decoyReg = new Regex(defaultDecoyPattern);
IIdentifiedProteinGroupFilter decoyFilter = null;
IFalseDiscoveryRateCalculator curCalc = null;
var paramFile = FileUtils.ChangeExtension(proteinFile, ".param");
if (File.Exists(paramFile))
{
BuildSummaryOptions options = BuildSummaryOptionsUtils.LoadFromFile(paramFile);
if (options.FalseDiscoveryRate.FilterByFdr)
{
decoyFilter = options.GetDecoyGroupFilter();
curCalc = options.FalseDiscoveryRate.GetFalseDiscoveryRateCalculator();
}
}
if (decoyFilter == null)
{
decoyFilter = new IdentifiedProteinGroupNameRegexFilter(defaultDecoyPattern, false);
curCalc = defaultCalc;
}
var peptideFile = FileUtils.ChangeExtension(proteinFile, ".peptides");
if (File.Exists(peptideFile))
{
var peptides = new MascotPeptideTextFormat().ReadFromFile(peptideFile);
var fullSpectra = GetSpectraByNPT(peptides, 2);
var fullTargetSpectra = GetTargetSpectra(decoyReg, fullSpectra);
var semiSpectra = GetSpectraByNPT(peptides, 1);
var semiTargetSpectra = GetTargetSpectra(decoyReg, semiSpectra);
result.FullSpectrumCount = GetSpectrumCount(fullSpectra);
result.FullTargetSpectrumCount = GetSpectrumCount(fullTargetSpectra);
result.SemiSpectrumCount = GetSpectrumCount(semiSpectra);
result.SemiTargetSpectrumCount = GetSpectrumCount(semiTargetSpectra);
result.FullPeptideCount = IdentifiedSpectrumUtils.GetUniquePeptideCount(fullSpectra);
result.FullTargetPeptideCount = IdentifiedSpectrumUtils.GetUniquePeptideCount(fullTargetSpectra);
result.SemiPeptideCount = IdentifiedSpectrumUtils.GetUniquePeptideCount(semiSpectra);
result.SemiTargetPeptideCount = IdentifiedSpectrumUtils.GetUniquePeptideCount(semiTargetSpectra);
result.FullSpectrumFdr = curCalc.Calculate(result.FullSpectrumCount - result.FullTargetSpectrumCount, result.FullTargetSpectrumCount);
result.SemiSpectrumFdr = curCalc.Calculate(result.SemiSpectrumCount - result.SemiTargetSpectrumCount, result.SemiTargetSpectrumCount);
result.FullPeptideFdr = curCalc.Calculate(result.FullPeptideCount - result.FullTargetPeptideCount, result.FullTargetPeptideCount);
result.SemiPeptideFdr = curCalc.Calculate(result.SemiPeptideCount - result.SemiTargetPeptideCount, result.SemiTargetPeptideCount);
}
if (File.Exists(proteinFile))
{
var ir = new MascotResultTextFormat().ReadFromFile(proteinFile);
ir.InitUniquePeptideCount();
var u2proteins = (from p in ir
where p[0].UniquePeptideCount > 1
select p).ToList();
var u1proteins = (from p in ir
where p[0].UniquePeptideCount == 1
select p).ToList();
result.ProteinGroupCount = ir.Count;
result.Unique2ProteinGroupCount = u2proteins.Count;
int targetCount;
result.Unique2ProteinFdr = CalculateProteinFdr(u2proteins, decoyFilter, defaultCalc, out targetCount);
result.Unique2ProteinGroupTargetCount = (int)targetCount;
result.Unique1ProteinFdr = CalculateProteinFdr(u1proteins, decoyFilter, defaultCalc, out targetCount);
result.Unique1ProteinGroupTargetCount = (int)targetCount;
}
return(result);
}