Updated PEP to use set seed (42)

MetaMorpheus/EngineLayer/FdrAnalysis/FdrInfo.cs

@@ -10,5 +10,16 @@ public class FdrInfo
         public double QValueNotch { get; set; }
         public double PEP { get; set; }
         public double PEP_QValue { get; set; }
+
+        /// <summary>
+        /// Creates a new FdrInfo object where Q-Values and PEP_Qvalues are set to 2 by default
+        /// This is done to avoid situations where q-values aren't calcualted for a given peptides, but it is still
+        /// reported in the final results. 
+        /// </summary>
+        public FdrInfo()
+        {
+            QValue = 2;
+            PEP_QValue = 2;
+        }
     }
 }

MetaMorpheus/EngineLayer/FdrAnalysis/PEPAnalysisEngine.cs

@@ -22,6 +22,27 @@ namespace EngineLayer
 {
     public class PepAnalysisEngine
     {
+        private int _randomSeed = 42;
+
+        /// <summary>
+        /// This method contains the hyper-parameters that will be used when training the machine learning model
+        /// </summary>
+        /// <returns> Options object to be passed in to the FastTree constructor </returns>
+        public Microsoft.ML.Trainers.FastTree.FastTreeBinaryTrainer.Options BGDTreeOptions =>
+            new Microsoft.ML.Trainers.FastTree.FastTreeBinaryTrainer.Options
+            {
+                NumberOfThreads = 1,
+                NumberOfTrees = 400,
+                MinimumExampleCountPerLeaf = 10,
+                NumberOfLeaves = 20,
+                LearningRate = 0.2,
+                LabelColumnName = "Label",
+                FeatureColumnName = "Features",
+                Seed = _randomSeed,
+                FeatureSelectionSeed = _randomSeed,
+                RandomStart = false
+            };
+
         private static readonly double AbsoluteProbabilityThatDistinguishesPeptides = 0.05;
 
         //These two dictionaries contain the average and standard deviations of hydrophobicitys measured in 1 minute increments accross each raw
@@ -105,10 +126,10 @@ public string ComputePEPValuesForAllPSMs()
                 }
             }
 
-            MLContext mlContext = new MLContext();
+            MLContext mlContext = new MLContext(seed: _randomSeed);
             TransformerChain<BinaryPredictionTransformer<Microsoft.ML.Calibrators.CalibratedModelParametersBase<Microsoft.ML.Trainers.FastTree.FastTreeBinaryModelParameters, Microsoft.ML.Calibrators.PlattCalibrator>>>[] trainedModels = new TransformerChain<BinaryPredictionTransformer<Microsoft.ML.Calibrators.CalibratedModelParametersBase<Microsoft.ML.Trainers.FastTree.FastTreeBinaryModelParameters, Microsoft.ML.Calibrators.PlattCalibrator>>>[numGroups];
 
-            var trainer = mlContext.BinaryClassification.Trainers.FastTree(labelColumnName: "Label", featureColumnName: "Features", numberOfTrees: 400);
+            var trainer = mlContext.BinaryClassification.Trainers.FastTree(BGDTreeOptions);
             var pipeline = mlContext.Transforms.Concatenate("Features", TrainingVariables)
                 .Append(trainer);
 

MetaMorpheus/EngineLayer/PeptideSpectralMatch.cs

@@ -47,6 +47,10 @@ protected PeptideSpectralMatch(SpectralMatch psm, List<(int Notch, IBioPolymerWi
             : base(psm, bestMatchingPeptides)
         {
         }
+
+        public PeptideSpectralMatch(SpectralMatch psm) : base(psm)
+        {
+        }
         #endregion
 
 

MetaMorpheus/EngineLayer/SpectralMatch.cs

@@ -379,6 +379,46 @@ protected SpectralMatch(SpectralMatch psm, List<(int Notch, IBioPolymerWithSetMo
             SpectralAngle = psm.SpectralAngle;
         }
 
+        public SpectralMatch(SpectralMatch psm)
+        {
+            _BestMatchingBioPolymersWithSetMods = psm._BestMatchingBioPolymersWithSetMods;
+            BaseSequence = PsmTsvWriter.Resolve(_BestMatchingBioPolymersWithSetMods.Select(b => b.Pwsm.BaseSequence)).ResolvedValue;
+            FullSequence = PsmTsvWriter.Resolve(_BestMatchingBioPolymersWithSetMods.Select(b => b.Pwsm.FullSequence)).ResolvedValue;
+
+            ModsChemicalFormula = psm.ModsChemicalFormula;
+            Notch = psm.Notch;
+            BioPolymerWithSetModsLength = psm.BioPolymerWithSetModsLength;
+            OneBasedStartResidue = psm.OneBasedStartResidue;
+            OneBasedEndResidue = psm.OneBasedEndResidue;
+            BioPolymerWithSetModsMonoisotopicMass = psm.BioPolymerWithSetModsMonoisotopicMass;
+            ParentLength = psm.ParentLength;
+            Accession = psm.Accession;
+            Organism = psm.Organism;
+            MatchedFragmentIons = psm.MatchedFragmentIons;
+            PsmCount = psm.PsmCount;
+            ModsIdentified = psm.ModsIdentified;
+            LocalizedScores = psm.LocalizedScores;
+            ScanNumber = psm.ScanNumber;
+            PrecursorScanNumber = psm.PrecursorScanNumber;
+            ScanRetentionTime = psm.ScanRetentionTime;
+            ScanExperimentalPeaks = psm.ScanExperimentalPeaks;
+            TotalIonCurrent = psm.TotalIonCurrent;
+            ScanPrecursorCharge = psm.ScanPrecursorCharge;
+            ScanPrecursorMonoisotopicPeakMz = psm.ScanPrecursorMonoisotopicPeakMz;
+            ScanPrecursorMass = psm.ScanPrecursorMass;
+            FullFilePath = psm.FullFilePath;
+            ScanIndex = psm.ScanIndex;
+            FdrInfo = psm.FdrInfo;
+            Score = psm.Score;
+            Xcorr = psm.Xcorr;
+            RunnerUpScore = psm.RunnerUpScore;
+            IsDecoy = psm.IsDecoy;
+            IsContaminant = psm.IsContaminant;
+            DigestionParams = psm.DigestionParams;
+            BioPolymersWithSetModsToMatchingFragments = psm.BioPolymersWithSetModsToMatchingFragments;
+            SpectralAngle = psm.SpectralAngle;
+        }
+
         #endregion
 
         #region FDR

MetaMorpheus/TaskLayer/SearchTask/PostSearchAnalysisTask.cs

@@ -137,10 +137,37 @@
             // this could cause weird PSM FDR issues
 
             Status("Estimating PSM FDR...", Parameters.SearchTaskId);
-            new FdrAnalysisEngine(psms, Parameters.NumNotches, CommonParameters, this.FileSpecificParameters, 
-                new List<string> { Parameters.SearchTaskId }, analysisType: analysisType, doPEP: doPep, outputFolder: Parameters.OutputFolder).Run();
 
+            List<int> psmsAboveQ = new List<int>();
+            List<int> peptidesAboveQ = new List<int>();
+            List<int> psmsAbovePepQ = new List<int>();
+            List<int> peptidesAbovePepQ = new List<int>();
+
+            for (int i = 0; i < 10; i++)
+            {
+                var tempPsms = psms
+                    .Select(p => new PeptideSpectralMatch(p))
+                    .Cast<SpectralMatch>()
+                    .ToList(); 
+                new FdrAnalysisEngine(tempPsms, Parameters.NumNotches, CommonParameters, this.FileSpecificParameters,
+                    new List<string> { Parameters.SearchTaskId }, analysisType: analysisType, doPEP: doPep, outputFolder: Parameters.OutputFolder).Run();
+                psmsAboveQ.Add(tempPsms.Count(p => p.FdrInfo.QValue <= 0.01));
+                peptidesAboveQ.Add(tempPsms.Count(p => p.PeptideFdrInfo.QValue <= 0.01));
+                psmsAbovePepQ.Add(tempPsms.Count(p => p.FdrInfo.PEP_QValue <= 0.01));
+                peptidesAbovePepQ.Add(tempPsms.Count(p => p.PeptideFdrInfo.PEP_QValue <= 0.01));
+            }
+
+            // write summary text
+            using(StreamWriter writer = new StreamWriter(@"C:\Users\Alex\PEP_NoRandomOrdering.tsv"))
+            {
+                writer.WriteLine("PSMs Q\tPeptides Q\tPSMs PEP-Q\tPeptides PEP-Q");
+                for(int i = 0; i < 10; i++)
+                {
+                    writer.WriteLine(String.Join('\t', psmsAboveQ[i], peptidesAboveQ[i], psmsAbovePepQ[i], peptidesAbovePepQ[i]));
+                }
+            }
+            throw new Exception("PEP Seed Results written to file");
             Status("Done estimating PSM FDR!", Parameters.SearchTaskId);
         }
 
         private void ProteinAnalysis()