private static List <PsmTsvLine> AssignFDRToTarget(string[] primaryLines, string[] secondaryLines)
{
List <PsmTsvLine> primaryPsms = ImportPsmtsv.ImportLinesToAggregate(primaryLines);
List <PsmTsvLine> secondaryPsms = ImportPsmtsv.ImportLinesToAggregate(secondaryLines);
primaryPsms = primaryPsms.OrderByDescending(x => x.score).ToList();
secondaryPsms = secondaryPsms.OrderByDescending(x => x.score).ToList();
int p = 0;
int s = 0;
int target = 0;
int decoy = 0;
double qMax = 0;
PsmTsvLine decoyLine = secondaryPsms[s];
while (p < primaryPsms.Count)
{
PsmTsvLine targetLine = primaryPsms[p];
if (targetLine.score > decoyLine.score || s == secondaryPsms.Count)
{
target++;
targetLine.target = target.ToString();
targetLine.decoy = decoy.ToString();
double qValue = (1.0d * decoy / target);
qMax = (qMax > qValue) ? qMax : qValue;
targetLine.q = qMax.ToString();
p++;
}
else
{
decoy++;
s++;
}
}
return(primaryPsms);
}
public PsmTsvLine AggregateLine(PsmTsvLine secondary)
{
baseSequence += "|" + secondary.baseSequence;
fullSequence += "|" + secondary.fullSequence;
accession += "|" + secondary.accession;
proteinName += "|" + secondary.proteinName;
geneName += "|" + secondary.geneName;
return(this);
}
private static List <PsmTsvLine> AggregateDifferentDatabaseSearches(string[] primaryLines, string[] secondaryLines)
{
List <PsmTsvLine> primaryPsms = ImportPsmtsv.ImportLinesToAggregate(primaryLines);
List <PsmTsvLine> secondaryPsms = ImportPsmtsv.ImportLinesToAggregate(secondaryLines);
List <PsmTsvLine> aggregatedLines = new List <PsmTsvLine>();
int p = 0;
int s = 0;
while (p < primaryPsms.Count && s < secondaryPsms.Count)
{
PsmTsvLine psmP = primaryPsms[p];
PsmTsvLine psmS = secondaryPsms[s];
if (psmP.scanNumber < psmS.scanNumber)
{
aggregatedLines.Add(psmP);
p++;
}
else if (psmP.scanNumber > psmS.scanNumber)
{
aggregatedLines.Add(psmS);
s++;
}
else
{
if (psmP.score > psmS.score - 0.001 && psmP.score < psmS.score + 0.001)
{
aggregatedLines.Add(psmP.AggregateLine(psmS));
}
else if (psmP.score > psmS.score)
{
aggregatedLines.Add(psmP);
}
else
{
aggregatedLines.Add(psmS);
}
p++;
s++;
}
}
while (p < primaryPsms.Count)
{
aggregatedLines.Add(primaryPsms[p]);
p++;
}
while (s < secondaryPsms.Count)
{
aggregatedLines.Add(primaryPsms[s]);
s++;
}
return(aggregatedLines);
}
public static List <PsmTsvLine> Percolate(List <PsmTsvLine> primaryPsms, List <PsmTsvLine> secondaryPsms, double qThreshold, double scoreDifferenceThreshold)
{
List <PsmTsvLine> aggregatedLines = new List <PsmTsvLine>();
int p = 0;
int s = 0;
//While loop generates a combined list of primary (normal) and secondary (spliced) psms based on scoreDifferenceThreshold
while (p < primaryPsms.Count && s < secondaryPsms.Count)
{
PsmTsvLine psmP = primaryPsms[p];
PsmTsvLine psmS = secondaryPsms[s];
if (psmP.scanNumber < psmS.scanNumber)
{
aggregatedLines.Add(psmP);
p++;
}
else if (psmP.scanNumber > psmS.scanNumber)
{
aggregatedLines.Add(psmS);
s++;
}
else
{
if (psmP.score > psmS.score - scoreDifferenceThreshold)
{
aggregatedLines.Add(psmP);
}
else
{
psmS.neoType = (Convert.ToDouble(psmP.q) < qThreshold) ? PsmTsvLine.NeoType.DecoySpliced : PsmTsvLine.NeoType.Spliced; //if the beaten score belonged to a gold standard, this is a false discovery
aggregatedLines.Add(psmS);
}
p++;
s++;
}
}
//wrap up any leftover psms without scanCounts
for (; p < primaryPsms.Count; p++)
{
aggregatedLines.Add(primaryPsms[p]);
}
for (; s < secondaryPsms.Count; s++)
{
aggregatedLines.Add(secondaryPsms[s]);
}
return(aggregatedLines);
}