public double ComputeFDR(PeptideSpectrumMatches matches, out long nbTarget, out long nbDecoy)
{
nbTarget = 0;
nbDecoy = 0;
foreach (PeptideSpectrumMatch psm in matches)
{
if (KeepPSM(psm))
{
if (psm.Target)
{
nbTarget++;
}
else
{
nbDecoy++;
}
}
}
if (nbTarget == 0)
{
return(1);
}
else
{
return((double)nbDecoy / (double)nbTarget);
}
}
public Precursor()
{
this.psms_AllPossibilities = new PeptideSpectrumMatches(); // GraphML_List<PeptideSpectrumMatch>();
this.psms = new PeptideSpectrumMatches(); // GraphML_List<PeptideSpectrumMatch>();
Isotopes = new GraphML_List <Precursor>();
OtherCharges = new GraphML_List <Precursor>();
}
public Precursor(Track track, int charge, Sample entry, double massShift = 0, GraphML_List <Precursor> isotopes = null)//double mz, double intensity, int charge = -1, double rt = 0, double massShift = 0, List<Precursor> isotopes = null)
{
INDEX = COMPTEUR++;
this.Track = track;
this.Charge = charge;
this.sample = entry;
this.psms_AllPossibilities = new PeptideSpectrumMatches(); // GraphML_List<PeptideSpectrumMatch>();
this.psms = new PeptideSpectrumMatches(); // GraphML_List<PeptideSpectrumMatch>();
// this.Mz = mz;
// this.Intensity = intensity;
// this.Charge = charge;
this.Mass = Numerics.MassFromMZ(track.MZ, charge);
// this.Rt = rt;
MassShift = massShift;
if (isotopes == null)
{
Isotopes = new GraphML_List <Precursor>();
}
else
{
Isotopes = isotopes;
}
OtherCharges = new GraphML_List <Precursor>();
}
/// <summary>
/// Aligns fragments observed Mz values
/// </summary>
/// <param name="result"></param>
/// <param name="allPSMs"></param>
public static void AlignProductsByDiff(Result result, PeptideSpectrumMatches allPSMs)
{
List <double> observedDiff = new List <double>();
List <double> observedMz = new List <double>();
//Precursors and Tracks
foreach (Precursor precursor in result.matchedPrecursors)//.ComputeAtFDR(result.dbOptions.maximumFalseDiscoveryRate, false))
{
PeptideSpectrumMatch psm = precursor.OptimizedBestPsm();
if (psm.Target)
{
foreach (ProductMatch fragment in psm.AllProductMatches)
{
observedMz.Add(fragment.obsMz);
observedDiff.Add(fragment.mass_diff);
}
}
}
PolynominalRegression pr = new PolynominalRegression(observedMz, observedDiff, 2);
foreach (PeptideSpectrumMatch psm in allPSMs)
{
foreach (ProductMatch match in psm.AllProductMatches)
{
match.obsMz += pr.Calculate(match.obsMz);
match.mass_diff = match.theoMz - match.obsMz;
}
psm.Initialize(result.dbOptions, psm.AllProductMatches);
foreach (MsMsPeak peak in psm.Query.spectrum.Peaks)
{
peak.MZ += pr.Calculate(peak.MZ);
}
}
}
public Query(DBOptions dbOptions, Sample entry, ProductSpectrum spectrum, Precursor precursor, int spectraIndex = -1)
{
this.options = dbOptions;
this.sample = entry;
this.psms = new PeptideSpectrumMatches();
this.spectrum = spectrum;
this.precursor = precursor;
this.spectrumIndex = spectraIndex;
if (spectrum != null && precursor != null)
{
this.TrapDistance = Math.Abs(spectrum.PrecursorMZ - precursor.Track.MZ);
}
}
/// <summary>
/// Crop observed precursors values outside the variance or Standard Deviation (whichever is bigger)
/// </summary>
/// <param name="result"></param>
/// <param name="allPSMs"></param>
/// <returns>Returns the newly computed Precursor tolerance</returns>
public static double CropPrecursors(Result result, PeptideSpectrumMatches allPSMs)
{
List <double> errorPrecursor = new List <double>(result.precursors.Count);
foreach (Precursor precursor in result.matchedPrecursors)
{
PeptideSpectrumMatch psm = precursor.OptimizedBestPsm();
if (psm.Target)
{
errorPrecursor.Add(psm.PrecursorMzError);
}
}
double variance = Numerics.Variance(errorPrecursor);
double stdev = Numerics.StandardDeviation(errorPrecursor);
result.dbOptions.ConSole.WriteLine("Computed Precursor Variance = " + variance + " STDev = " + stdev);
if (variance < stdev)
{
variance = stdev;
}
//variance = result.dbOptions.precursorMassTolerance.Value * ((2 * variance) / result.dbOptions.precursorMassTolerance.Value);
int nbRemovedPSM = 0;
foreach (Precursor precursor in result.precursors)
{
for (int i = 0; i < precursor.psms.Count;)
{
if (Math.Abs(precursor.psms[i].PrecursorMzError) > variance)
{
//allPSMs[i].Query.precursor.psms_AllPossibilities.Remove(allPSMs[i]);
precursor.psms.RemoveAt(i);
//allPSMs.RemoveAt(i);
nbRemovedPSM++;
}
else
{
i++;
}
}
}
result.dbOptions.ConSole.WriteLine("Removed " + nbRemovedPSM + " [" + allPSMs.Count + " remaining] Peptide Spectrum matches outside the variance [" + variance + "]");
return(variance);
}
/// <summary>
/// Aligns precursors observed Mz values
/// </summary>
/// <param name="result"></param>
/// <param name="allPSMs"></param>
public static void AlignPrecursorsByDiff(Result result, PeptideSpectrumMatches allPSMs)
{
List <double> observedDiff = new List <double>();
List <double> observedMz = new List <double>();
//Precursors and Tracks
foreach (Precursor precursor in result.matchedPrecursors)//.ComputeAtFDR(result.dbOptions.maximumFalseDiscoveryRate, false))
{
PeptideSpectrumMatch psm = precursor.OptimizedBestPsm();
if (psm.Target)
{
observedMz.Add(precursor.Track.MZ);
observedDiff.Add(Numerics.MZFromMass(psm.Peptide.MonoisotopicMass, precursor.Charge) - precursor.Track.MZ);
}
}
PolynominalRegression pr = new PolynominalRegression(observedMz, observedDiff, 2);
foreach (Query query in result.queries)
{
query.precursor.Track.MZ += pr.Calculate(query.precursor.Track.MZ);
query.precursor.Mass = Numerics.MassFromMZ(query.precursor.Track.MZ, query.precursor.Charge);
foreach (Precursor precursor in query.precursor.Isotopes)
{
precursor.Track.MZ += pr.Calculate(precursor.Track.MZ);
precursor.Mass = Numerics.MassFromMZ(precursor.Track.MZ, precursor.Charge);
}
foreach (Precursor precursor in query.precursor.OtherCharges)
{
precursor.Track.MZ += pr.Calculate(precursor.Track.MZ);
precursor.Mass = Numerics.MassFromMZ(precursor.Track.MZ, precursor.Charge);
}
query.spectrum.PrecursorMZ += pr.Calculate(query.spectrum.PrecursorMZ);
query.spectrum.PrecursorMass = Numerics.MassFromMZ(query.spectrum.PrecursorMZ, query.spectrum.PrecursorCharge);
}
foreach (PeptideSpectrumMatch psm in allPSMs)
{
psm.UpdatePrecursor(result.dbOptions);
}
}
}//*/
/// <summary>
/// Latest version of the search routine. Associates spectrum to digested peptide sequences,
/// aligns precursors and fragments, clusters common precursors accross samples, create the list of detected
/// peptide and protein sequences.
/// TODO Align maps together (retention time)
/// </summary>
/// <param name="queries"></param>
/// <returns></returns>
public Result SearchVersionAugust2013(Queries queries, bool optimize)
{
Result result = new Result();
result.queries = queries;
result.dbOptions = dbOptions;
result.samples = Project;
DBSearcher dbSearcher = new DBSearcher(dbOptions);
Digestion ps = new Digestion(dbOptions);
if (dbOptions.NoEnzymeSearch)
{
result.SetPrecursors(dbSearcher.Search(queries, ps.DigestProteomeOnTheFlyNoEnzyme(AllProteins, queries)));
}
else
{
result.SetPrecursors(dbSearcher.Search(queries, ps.DigestProteomeOnTheFly(AllProteins, false, queries)));
}
dbOptions.ConSole.WriteLine(result.precursors.Count + " precursors matched !");
PeptideSpectrumMatches allPSMs = new PeptideSpectrumMatches();
foreach (Precursor precursor in result.precursors)
{
foreach (PeptideSpectrumMatch psm in precursor.psms_AllPossibilities)
{
allPSMs.Add(psm);
}
}
//Add all psm possibilities to psms list
foreach (PeptideSpectrumMatch psm in allPSMs)
{
psm.Query.precursor.psms.Add(psm);
}
long nbTargets = result.SetPrecursors(result.precursors);
dbOptions.ConSole.WriteLine("Targets before Optimizing Score Ratios : " + nbTargets + " [" + result.matchedPrecursors.Count + "]");
if (optimize)
{
//allPSMs.OptimizePSMScoreRatios(dbOptions, dbOptions.PSMFalseDiscoveryRate, result);
//result.matchedPrecursors.OptimizePSMScoreRatios(dbOptions, dbOptions.PSMFalseDiscoveryRate, result);
nbTargets = result.SetPrecursors(result.precursors);
dbOptions.ConSole.WriteLine("Targets : " + nbTargets + " [" + result.matchedPrecursors.Count + "]");
//*/
//TODO Improve alignment results
/*
* Align.AlignPrecursorsByDiff(result, allPSMs);
* nbTargets = result.SetPrecursors(result.precursors);
* dbOptions.ConSole.WriteLine("Targets after precursor alignment : " + nbTargets + " [" + result.matchedPrecursors.Count + "]");
*
* Align.AlignProductsByDiff(result, allPSMs);
* nbTargets = result.SetPrecursors(result.precursors);
* dbOptions.ConSole.WriteLine("Targets after fragment alignment : " + nbTargets + " [" + result.matchedPrecursors.Count + "]");
* //*/
/*
* dbOptions.precursorMassTolerance.Value = Align.CropPrecursors(result, allPSMs);
* nbTargets = result.SetPrecursors(result.precursors);
* dbOptions.ConSole.WriteLine("Targets after croping precursors : " + nbTargets + " [" + result.matchedPrecursors.Count + "]");
*
* dbOptions.productMassTolerance.Value = Align.CropProducts(result, allPSMs);
* nbTargets = result.SetPrecursors(result.precursors);
* dbOptions.ConSole.WriteLine("Targets after croping fragments : " + nbTargets + " [" + result.matchedPrecursors.Count + "]");
* //*/
}
//*/
allPSMs = null;
long bestTargets = nbTargets;
MSSearcher msSearcher = new MSSearcher(dbOptions, Project);
msSearcher.CumulPsm(result.matchedPrecursors);//TODO Check if its still needed
//Step 1 : Cluster psms together based on precursor feature //TODO Implement ProteoProfile Scoring based clustering
//Group in clusters
result.clusters = msSearcher.Search(result.matchedPrecursors, true);
//Todo Align retention times
//Todo redo clusterization, based on retention time aligned maps
//Step 2 : Regroup based on peptide sequence (Morpheus code)
PeptideSearcher pepSearcher = new PeptideSearcher(dbOptions);
result.peptides = pepSearcher.Search(result.clusters, result.matchedPrecursors, true);
result.peptideSequences = pepSearcher.Search(result.clusters, result.matchedPrecursors, false);
//Step 3 : Regroup based on protein sequences (Morpheus code)
ProteinSearcher protSearcher = new ProteinSearcher(dbOptions);
result.proteins = protSearcher.SearchLatest(result.peptideSequences);//, dbSearcher.DicOfProteins);
//long lastNbTarget = nbTargets;
//do
//{
// lastNbTarget = nbTargets;
//UpdatePsmScores(result.proteins);
// nbTargets = result.SetPrecursors(result.precursors);
//} while (lastNbTarget < nbTargets);//*/
result.peptides = pepSearcher.Search(result.clusters, result.matchedPrecursors, true);
result.peptideSequences = pepSearcher.Search(result.clusters, result.matchedPrecursors, false);
if (optimize)
{
dbOptions.ConSole.WriteLine("Targets before second Optimization of Score Ratios : " + nbTargets + " [" + result.matchedPrecursors.Count + "]");
//result.matchedPrecursors.OptimizePSMScoreRatios(dbOptions, dbOptions.PSMFalseDiscoveryRate, result);
nbTargets = result.SetPrecursors(result.precursors);
dbOptions.ConSole.WriteLine("Targets after ReOptimizing PSM Score Ratios : " + nbTargets + " [" + result.matchedPrecursors.Count + "]");//*/
}
//Step 5 : Compute the new number of Targets
nbTargets = result.SetPrecursors(result.precursors);
if (nbTargets < bestTargets)
{
dbOptions.ConSole.WriteLine("FAILED to improve PSMs while adding protein and peptide information");
}
dbOptions.ConSole.WriteLine(result.matchedPrecursors.Count + " precursors remaining after ProPheus Search!");
return(result);
}//*/
public static double CropProductsBKP(Result result, PeptideSpectrumMatches allPSMs)
{
List <double> errorProduct = new List <double>(result.precursors.Count);
foreach (Precursor precursor in result.matchedPrecursors)
{
PeptideSpectrumMatch psm = precursor.OptimizedBestPsm();
if (psm.Target)
{
foreach (ProductMatch match in psm.AllProductMatches)
{
errorProduct.Add(match.mass_diff);
}
}
}
double variance = Numerics.Variance(errorProduct);
double stdev = Numerics.StandardDeviation(errorProduct);
result.dbOptions.ConSole.WriteLine("Computed Product Variance = " + variance + " STDev = " + stdev);
if (variance < stdev)
{
variance = stdev;
}
//variance = result.dbOptions.productMassTolerance.Value * ((2 * variance) / result.dbOptions.productMassTolerance.Value);
int nbRemovedProduct = 0;
foreach (PeptideSpectrumMatch psm in allPSMs)
{
for (int i = 0; i < psm.AllProductMatches.Count;)
{
if (Math.Abs(psm.AllProductMatches[i].mass_diff) > variance)
{
psm.AllProductMatches.RemoveAt(i);
nbRemovedProduct++;
}
else
{
i++;
}
}
psm.MatchingProducts = psm.AllProductMatches.Count;
}
int nbRemovedPSM = 0;
foreach (Precursor precursor in result.precursors)
{
for (int i = 0; i < precursor.psms.Count;)
{
if (precursor.psms[i].MatchingProducts < 2)
{
precursor.psms.RemoveAt(i);
}
else
{
precursor.psms[i].Initialize(result.dbOptions, precursor.psms[i].AllProductMatches);
i++;
}
}
}
result.dbOptions.ConSole.WriteLine("Removed " + nbRemovedProduct + " [" + nbRemovedPSM + " removed PSMs] Fragment matches outside the variance [" + variance + "]");
return(variance);
}
public Query()
{
this.psms = new PeptideSpectrumMatches();
}
