public List <ProteinSpectrumMatch> ReadMsGfPlusResult(string msgfResultPath, int maxPrsm)
{
var parser = new TsvFileParser(msgfResultPath);
var prsmList = new List <ProteinSpectrumMatch>();
var prevScanNum = -1;
for (var i = 0; i < parser.NumData; i++)
{
var sequence = parser.GetData("Peptide")[i];
var scanNum = int.Parse(parser.GetData("Scan")[i]);
if (prevScanNum == scanNum)
{
continue;
}
prevScanNum = scanNum;
var mz = double.Parse(parser.GetData("PrecursorMZ")[i]);
var protName = parser.GetData("Protein")[i];
var protDesc = "";
var score = double.Parse(parser.GetData("MSGFScore")[i]);
var charge = int.Parse(parser.GetData("Charge")[i]);
var seq = Sequence.GetSequenceFromMsGfPlusPeptideStr(sequence);
var sequenceText = GetSequenceText(seq);
var mass = (mz - Constants.Proton) * charge;
var firstResId = 0;
var lastResId = 0;
var fdr = Double.Parse(parser.GetData("QValue")[i]);
if (fdr > FdrCutoff)
{
continue;
}
var prsm = new ProteinSpectrumMatch(sequence, scanNum, mass, charge, protName, protDesc, firstResId, lastResId, score, ProteinSpectrumMatch.SearchTool.MsGfPlus)
{
SequenceText = sequenceText,
};
prsmList.Add(prsm);
if (prsmList.Count >= maxPrsm)
{
break;
}
}
return(prsmList);
}
public List <ProteinSpectrumMatch> ReadMsAlignResult(string msAlignResultTablePath, int maxPrsm)
{
var parser = new TsvFileParser(msAlignResultTablePath);
var prsmList = new List <ProteinSpectrumMatch>();
for (var i = 0; i < parser.NumData; i++)
{
var sequence = parser.GetData("Peptide")[i];
var scanNum = int.Parse(parser.GetData("Scan(s)")[i]);
var mass = double.Parse(parser.GetData("Precursor_mass")[i]);
var protNameDesc = parser.GetData("Protein_name")[i];
var k = protNameDesc.IndexOf(' ');
var protName = (k < 0) ? protNameDesc : protNameDesc.Substring(0, k);
var protDesc = (k < 0) ? protNameDesc : protNameDesc.Substring(k + 1);
var firstResId = int.Parse(parser.GetData("First_residue")[i]);
var lastResId = int.Parse(parser.GetData("Last_residue")[i]);
var score = double.Parse(parser.GetData("#matched_fragment_ions")[i]);
var sequenceText = parser.GetData("Peptide")[i];
var charge = int.Parse(parser.GetData("Charge")[i]);
var evalue = double.Parse(parser.GetData("E-value")[i]);
var fdr = Double.Parse(parser.GetData("FDR")[i]);
if (fdr > FdrCutoff)
{
continue;
}
var prsm = new ProteinSpectrumMatch(sequence, scanNum, mass, charge, protName, protDesc, firstResId, lastResId, score, ProteinSpectrumMatch.SearchTool.MsAlign)
{
SequenceText = sequenceText,
SpectralEvalue = evalue,
};
prsmList.Add(prsm);
if (prsmList.Count >= maxPrsm)
{
break;
}
}
return(prsmList);
}
public List <ProteinSpectrumMatch> ReadMsPathFinderResult(string msPathFinderResultPath, int maxPrsm, double minScore = 3, double maxScore = int.MaxValue)
{
var prsmList = new List <ProteinSpectrumMatch>();
foreach (var result in DatabaseSearchResultData.ReadResultsFromFile(msPathFinderResultPath))
{
if (result.NumMatchedFragments < minScore || maxScore < result.NumMatchedFragments)
{
continue;
}
if (result.HasTdaScores && result.QValue > FdrCutoff)
{
continue;
}
var prsm = new ProteinSpectrumMatch(result.Sequence, result.ScanNum, result.Mass, result.Charge, result.ProteinName, result.ProteinDescription, result.Start, result.End, result.NumMatchedFragments, ProteinSpectrumMatch.SearchTool.MsPathFinder)
{
SequenceText = GetSequenceText(result.Sequence, result.Modifications),
Modifications = result.Modifications,
Pre = result.Pre,
Post = result.Post,
ProteinLength = result.ProteinLength,
SpectralEvalue = result.SpecEValue,
};
prsmList.Add(prsm);
if (prsmList.Count >= maxPrsm)
{
break;
}
}
return(prsmList);
}
public List <ProteinSpectrumMatch> ReadMsPathFinderResultOld(string msPathFinderResultPath, int maxPrsm, double minScore = 3, double maxScore = int.MaxValue)
{
var parser = new TsvFileParser(msPathFinderResultPath);
var prsmList = new List <ProteinSpectrumMatch>();
var scoreColumn = parser.GetData("#MatchedFragments") ?? parser.GetData("Score");
var qValColumn = parser.GetData("QValue");
var evalueColumn = parser.GetData("SpecEValue");
for (var i = 0; i < parser.NumData; i++)
{
var sequence = parser.GetData("Sequence")[i];
var scanNum = int.Parse(parser.GetData("Scan")[i]);
var mass = double.Parse(parser.GetData("Mass")[i]);
var protName = parser.GetData("ProteinName")[i];
var protDesc = parser.GetData("ProteinDesc")[i];
var charge = int.Parse(parser.GetData("Charge")[i]);
var firstResId = int.Parse(parser.GetData("Start")[i]);
var lastResId = int.Parse(parser.GetData("End")[i]);
var score = double.Parse(scoreColumn[i]);
var mod = parser.GetData("Modifications")[i];
var evalue = (evalueColumn != null) ? double.Parse(parser.GetData("SpecEValue")[i]) : 0;
var pre = parser.GetData("Pre")[i];
var post = parser.GetData("Post")[i];
var proteinLen = int.Parse(parser.GetData("ProteinLength")[i]);
if (score < minScore || score > maxScore)
{
continue;
}
if (qValColumn != null)
{
var fdr = double.Parse(qValColumn[i]);
if (fdr > FdrCutoff)
{
continue;
}
}
var sequenceText = GetSequenceText(sequence, mod);
var prsm = new ProteinSpectrumMatch(sequence, scanNum, mass, charge, protName, protDesc, firstResId, lastResId, score, ProteinSpectrumMatch.SearchTool.MsPathFinder)
{
SequenceText = sequenceText,
Modifications = mod,
Pre = pre,
Post = post,
ProteinLength = proteinLen,
SpectralEvalue = evalue,
};
prsmList.Add(prsm);
if (prsmList.Count >= maxPrsm)
{
break;
}
}
return(prsmList);
}