public MsGfMatch(string resultStr, MsGfPlusHeaderInformation header)
{
var token = resultStr.Split('\t');
if (token.Length != header.NumColumns)
{
IsValid = false;
}
else
{
SpecFile = token[header.SpecFileColNum];
Peptide = token[header.PeptideColNum];
if (header.FormulaColNum > 0)
{
Formula = Composition.Parse(token[header.FormulaColNum]);
}
ScanNum = Convert.ToInt32(token[header.ScanNumColNum]);
Charge = Convert.ToInt32(token[header.ChargeColNum]);
Protein = token[header.ProteinColNum];
DeNovoScore = Convert.ToInt32(token[header.DeNovoScoreColNum]);
MsgfScore = Convert.ToInt32(token[header.MsgfScoreColNum]);
SpecEValue = Convert.ToDouble(token[header.SpecEValueColNum]);
if (header.QValueColNum > 0)
{
QValue = Convert.ToDouble(token[header.QValueColNum]);
}
if (header.PepQValueColNum > 0)
{
PepQValue = Convert.ToDouble(token[header.PepQValueColNum]);
}
IsValid = true;
}
}
public MsGfResults(string resultFilePath)
{
_msGfMatches = new List <MsGfMatch>();
MsGfPlusHeaderInformation headerInfo = null;
var prevScanNum = -1;
foreach (var line in File.ReadLines(resultFilePath))
{
if (headerInfo == null && line.StartsWith("#"))
{
headerInfo = new MsGfPlusHeaderInformation(line);
continue;
}
var match = new MsGfMatch(line, headerInfo);
if (match.ScanNum == prevScanNum)
{
continue;
}
prevScanNum = match.ScanNum;
if (!match.IsValid || match.Protein.StartsWith(FastaDatabase.DecoyProteinPrefix))
{
continue;
}
_msGfMatches.Add(match);
}
_msGfMatches.Sort();
}
public int PostProcessing(string outputFilePath)
{
// Parse MS-GF+ results
var pepToResults = new Dictionary <string, MsGfMatch>();
MsGfPlusHeaderInformation headerInfo = null;
foreach (var line in File.ReadLines(MsGfResultPath))
{
if (line.StartsWith("#"))
{
headerInfo = new MsGfPlusHeaderInformation(line);
continue;
}
var match = new MsGfMatch(line, headerInfo);
if (!match.IsValid)
{
continue;
}
if (match.SpecEValue > SpecEValueThreshold)
{
continue;
}
if (!IsValid(match))
{
continue;
}
MsGfMatch prevMatch;
if (!pepToResults.TryGetValue(match.Peptide, out prevMatch))
{
pepToResults[match.Peptide] = match;
match.NumMatches = 1;
}
else
{
if (match.SpecEValue < prevMatch.SpecEValue)
{
pepToResults[match.Peptide] = match;
match.NumMatches += prevMatch.NumMatches;
}
else
{
++prevMatch.NumMatches;
}
}
}
//var filteredPsms = pepToResults.Select(entry => entry.Value).Where(IsValid).ToList();
var filteredPsms = pepToResults.Select(entry => entry.Value).ToList();
filteredPsms.Sort();
// compute FDR
var qValue = GetQValues(filteredPsms);
var index = -1;
var numId = 0;
using (var writer = new StreamWriter(outputFilePath))
{
writer.WriteLine("#SpecFile\tPeptide\tScanNum\tPrecursorMz\tCharge\tProtein\tNumMatches\tDeNovoScore\tMSGFScore\tSpecEValue\tPepQValue");
foreach (var match in filteredPsms)
{
//if (match.Protein.StartsWith("DecoyPrefix")) continue;
writer.WriteLine("{0}\t{1}\t{2}\t{3}\t{4}\t{5}\t{6}\t{7}\t{8}\t{9}\t{10}",
match.SpecFile
//, match.Peptide.Replace("C+57.021", "C")
, match.Peptide
, match.ScanNum
, new Ion(match.Formula, match.Charge).GetMonoIsotopicMz()
, match.Charge
, match.Protein
, match.NumMatches
, match.DeNovoScore
, match.MsgfScore
, match.SpecEValue
, qValue[++index]
);
if (!match.Protein.StartsWith(DecoyPrefix))
{
if (qValue[index] <= 0.01)
{
++numId;
}
}
}
}
return(numId);
}