public static void FdrThresholding(string[] rawFiles, string decoyPrefix, double peptideFdr, double peptidePep,
MascotQueryType type, bool perFileThreshold, bool debug, HashSet <string> sequences,
bool onlyCollectSequences, IProteinSet proteinSet,
HashSet <string> labelModificationSet, IIdentificationProvider ip)
{
CalcFdr(rawFiles, decoyPrefix, type, proteinSet, debug, labelModificationSet, ip);
double pepThreshVal;
double[] pepThreshVals = new double[rawFiles.Length];
if (peptideFdr < 1)
{
for (int i = 0; i < rawFiles.Length; i++)
{
pepThreshVals[i] = FindReverseHitThresholdValue(ip.GetIdentifications(rawFiles[i], type), peptideFdr,
decoyPrefix, proteinSet);
ip.Dispose();
if (peptidePep < 1)
{
pepThreshVals[i] = Math.Min(pepThreshVals[i], peptidePep);
}
}
pepThreshVal = ArrayUtil.Median(pepThreshVals);
if (peptidePep < 1)
{
pepThreshVal = Math.Min(pepThreshVal, peptidePep);
}
}
else
{
for (int i = 0; i < rawFiles.Length; i++)
{
pepThreshVals[i] = peptidePep;
}
pepThreshVal = peptidePep;
}
if (onlyCollectSequences)
{
for (int i = 0; i < rawFiles.Length; i++)
{
CollectSequences(ip.GetIdentifications(rawFiles[i], type),
perFileThreshold ? pepThreshVals[i] : pepThreshVal, sequences);
ip.Dispose();
}
}
else
{
for (int i = 0; i < rawFiles.Length; i++)
{
ApplyReverseHitThreshold(ip.GetIdentifications(rawFiles[i], type),
perFileThreshold ? pepThreshVals[i] : pepThreshVal);
ip.Dispose();
}
}
}
private static void LimitPep(string[] rawFiles, MascotQueryType type, IIdentificationProvider ip)
{
for (int i = 0; i < rawFiles.Length; i++)
{
Identifications ident = ip.GetIdentifications(rawFiles[i], type);
int n = ident.Count;
for (int j = 0; j < n; j++)
{
MascotPeptide p = ident.GetPeptidesAt(j)[0];
p.Pep = Math.Min(p.Pep, 1);
}
ident.Write();
ip.Dispose();
}
}
private static void CalcFdr(string[] rawFiles, string decoyPrefix, MascotQueryType type, IProteinSet proteinSet, bool debug,
IIdentificationProvider ip)
{
List <bool> correct = new List <bool>();
List <double> scores = new List <double>();
List <double> seqLen = new List <double>();
for (int i = 0; i < rawFiles.Length; i++)
{
Identifications ident = ip.GetIdentifications(rawFiles[i], type);
int n = ident.Count;
for (int j = 0; j < n; j++)
{
bool c = ident.IsHighestScoringCorrect(j, decoyPrefix, proteinSet);
double s = ident.GetHighestAltScore(j);
double l = Math.Log(ident.GetBestSequence(j).Length);
if (!double.IsNaN(s) && !double.IsInfinity(s))
{
correct.Add(c);
scores.Add(s);
seqLen.Add(l);
}
}
ip.Dispose();
if (correct.Count > 10000000)
{
break;
}
}
if (correct.Count == 0)
{
return;
}
bool write = debug && (type == MascotQueryType.Silac);
BayesianInversion2D bi = new BayesianInversion2D(scores.ToArray(), seqLen.ToArray(), correct.ToArray(), write);
if (write)
{
Write(rawFiles, bi);
}
for (int i = 0; i < rawFiles.Length; i++)
{
SetPep(ip.GetIdentifications(rawFiles[i], type), bi);
ip.Dispose();
}
}
public static void FdrThresholding(string[] rawFiles, string[] recalFiles, string[] nonRecalFiles, string revstring,
double peptideFdr, double peptidePep, bool perFileThreshold, bool keepLowScorers,
IProteinSet proteinSet, HashSet <string> labelModificationSet, IIdentificationProvider ip, bool writeOut)
{
HashSet <string> sequences = new HashSet <string>();
if (recalFiles.Length > 0)
{
FdrThresholding(recalFiles, revstring, peptideFdr, peptidePep, MascotQueryType.Silac,
perFileThreshold, writeOut, sequences, keepLowScorers, proteinSet,
labelModificationSet, ip);
FdrThresholding(recalFiles, revstring, peptideFdr, peptidePep, MascotQueryType.Isotope,
perFileThreshold, false, sequences, keepLowScorers, proteinSet,
labelModificationSet, ip);
}
if (nonRecalFiles.Length > 0)
{
FdrThresholding(nonRecalFiles, revstring, peptideFdr, peptidePep, MascotQueryType.Silac,
perFileThreshold, false, sequences, keepLowScorers, proteinSet,
labelModificationSet, ip);
FdrThresholding(nonRecalFiles, revstring, peptideFdr, peptidePep, MascotQueryType.Isotope,
perFileThreshold, false, sequences, keepLowScorers, proteinSet,
labelModificationSet, ip);
}
FdrThresholding(rawFiles, revstring, peptideFdr, peptidePep, MascotQueryType.Peak,
perFileThreshold,
false, sequences, keepLowScorers, proteinSet,
labelModificationSet, ip);
if (keepLowScorers)
{
FilterBySequence(rawFiles, revstring, peptideFdr, peptidePep, MascotQueryType.Silac,
perFileThreshold, writeOut, sequences, keepLowScorers, ip);
FilterBySequence(rawFiles, revstring, peptideFdr, peptidePep, MascotQueryType.Isotope,
perFileThreshold, false, sequences, keepLowScorers, ip);
FilterBySequence(rawFiles, revstring, peptideFdr, peptidePep, MascotQueryType.Peak,
perFileThreshold,
false, sequences, keepLowScorers, ip);
}
LimitPep(rawFiles, MascotQueryType.Silac, ip);
LimitPep(rawFiles, MascotQueryType.Isotope, ip);
LimitPep(rawFiles, MascotQueryType.Peak, ip);
}
public static void FilterBySequence(string[] rawFiles, string decoyPrefix, double totalPeptideFPR,
double specificPeptideFPR, MascotQueryType type, bool perFileThreshold, bool debug,
HashSet <string> sequences, bool onlyCollectSequences, IIdentificationProvider ip)
{
for (int i = 0; i < rawFiles.Length; i++)
{
FilterBySequence(ip.GetIdentifications(rawFiles[i], type), sequences);
ip.Dispose();
}
}
public static void CalcFdr(string[] rawFiles, string decoyPrefix, MascotQueryType type, IProteinSet proteinSet, bool debug,
HashSet <string> labelModificationSet, IIdentificationProvider ip)
{
CalcFdr(rawFiles, decoyPrefix, type, proteinSet, debug, ip);
}
