G-PTM-D Writes Decoys

MetaMorpheus/CMD/CMD.csproj

@@ -24,7 +24,7 @@
     <PackageReference Include="Microsoft.ML.CpuMath" Version="3.0.1" />
     <PackageReference Include="Microsoft.ML.FastTree" Version="3.0.1" />
     <PackageReference Include="Microsoft.NETCore.App" Version="2.2.8" />
-    <PackageReference Include="mzLib" Version="1.0.569" />
+    <PackageReference Include="mzLib" Version="5.3.1" />
     <PackageReference Include="Nett" Version="0.15.0" />
     <PackageReference Include="SQLite.Interop.dll" Version="1.0.103" />
     <PackageReference Include="System.Data.SQLite" Version="1.0.118" />

MetaMorpheus/CMD/Program.cs

@@ -9,6 +9,7 @@
 using System.IO;
 using System.Linq;
 using System.Text.RegularExpressions;
+using EngineLayer.DatabaseLoading;
 using Omics.Modifications;
 using TaskLayer;
 

MetaMorpheus/EngineLayer/DatabaseLoading/DbForTask.cs

@@ -0,0 +1,23 @@
+#nullable enable
+namespace EngineLayer.DatabaseLoading;
+
+public class DbForTask
+{
+    public DbForTask(string filePath, bool isContaminant, string? decoyIdentifier = null)
+    {
+        FilePath = filePath;
+        IsContaminant = isContaminant;
+        FileName = System.IO.Path.GetFileName(filePath);
+        IsSpectralLibrary = GlobalVariables.GetFileExtension(filePath).ToLowerInvariant() == ".msp";
+        DecoyIdentifier = decoyIdentifier ?? GlobalVariables.DecoyIdentifier;
+    }
+
+    public bool IsSpectralLibrary { get; }
+    public string FilePath { get; }
+    public bool IsContaminant { get; }
+    public string FileName { get; }
+    public string DecoyIdentifier { get; }
+    public int? BioPolymerCount { get; internal set; } = null;
+    public int? TargetCount { get; internal set; } = null;
+    public int? DecoyCount { get; internal set; } = null;
+}

MetaMorpheus/EngineLayer/EngineLayer.csproj

@@ -29,7 +29,7 @@
     <PackageReference Include="Microsoft.ML.CpuMath" Version="3.0.1" />
     <PackageReference Include="Microsoft.ML.FastTree" Version="3.0.1" />
     <PackageReference Include="Microsoft.NETCore.App" Version="2.2.8" />
-    <PackageReference Include="mzLib" Version="1.0.569" />
+    <PackageReference Include="mzLib" Version="5.3.1" />
     <PackageReference Include="NETStandard.Library" Version="2.0.3" />
     <PackageReference Include="Nett" Version="0.15.0" />
     <PackageReference Include="Newtonsoft.Json" Version="13.0.3" />

MetaMorpheus/EngineLayer/GlobalVariables.cs

@@ -21,6 +21,7 @@ namespace EngineLayer
 {
     public static class GlobalVariables
     {
+        public static string DecoyIdentifier { get; set; } = "DECOY";
         // for now, these are only used for error-checking in the command-line version.
         // compressed versions of the protein databases (e.g., .xml.gz) are also supported
         public static List<string> AcceptedDatabaseFormats { get; private set; }

MetaMorpheus/EngineLayer/Gptmd/GptmdEngine.cs

@@ -21,6 +21,7 @@ public class GptmdEngine : MetaMorpheusEngine
         private readonly Dictionary<string, Tolerance> FilePathToPrecursorMassTolerance; // this exists because of file-specific tolerances
         //The ScoreTolerance property is used to differentiatie when a PTM candidate is added to a peptide. We check the score at each position and then add that mod where the score is highest.
         private readonly double ScoreTolerance = 0.1;
+        private static readonly double QValueNotchThreshold = 0.05;
         public Dictionary<string, HashSet<Tuple<int, Modification>>> ModDictionary { get; init; }
         private readonly List<IGptmdFilter> Filters;
 
@@ -84,7 +85,9 @@ protected override MetaMorpheusEngineResults RunSpecific()
             int modsAdded = 0;
 
             int maxThreadsPerFile = CommonParameters.MaxThreadsToUsePerFile;
-            var psms = AllIdentifications.Where(b => b.FdrInfo.QValueNotch <= 0.05 && !b.IsDecoy).ToList();
+
+            // Keep psms if they are below q value notch threshold, or notch ambiguous results with hypothesis below threshold. 
+            var psms = AllIdentifications.Where(b => b.FdrInfo.QValueNotch <= QValueNotchThreshold || (Math.Abs(b.FdrInfo.QValueNotch - 2) < 0.001 && b.BestMatchingBioPolymersWithSetMods.Any(p => p.QValueNotch.HasValue && p.QValueNotch <= QValueNotchThreshold))).ToList();
             if (psms.Any() == false)
             {
                 return new GptmdResults(this, ModDictionary, 0);
@@ -106,9 +109,14 @@ protected override MetaMorpheusEngineResults RunSpecific()
                     var peptideTheorProducts = new List<Product>();
                     List<(int site, Modification mod, string proteinAccession)> bestMatches = [];
 
-                    foreach (var pepWithSetMods in psm.BestMatchingBioPolymersWithSetMods.Select(v => v.SpecificBioPolymer))
+                    foreach (var hypothesis in psm.BestMatchingBioPolymersWithSetMods.Select(v => v))
                     {
+                        // skip hypotheses that are above the q value notch threshold - will be null unless psm is notch ambiguous
+                        if (hypothesis.QValueNotch.HasValue && hypothesis.QValueNotch > QValueNotchThreshold)
+                            continue;
+
                         bestMatches.Clear();
+                        var pepWithSetMods = hypothesis.SpecificBioPolymer;
                         var isVariantProtein = pepWithSetMods.Parent != pepWithSetMods.Parent.ConsensusVariant;
                         var possibleModifications = GetPossibleMods(precursorMass, GptmdModifications, Combos,
                             FilePathToPrecursorMassTolerance[fileName], pepWithSetMods);

MetaMorpheus/EngineLayer/MetaMorpheusEngine.cs

@@ -6,6 +6,7 @@
 using System.Collections.Generic;
 using System.Diagnostics;
 using System.Linq;
+using System.Threading.Tasks;
 using Transcriptomics.Digestion;
 
 namespace EngineLayer
@@ -289,6 +290,8 @@ public MetaMorpheusEngineResults Run()
             return myResults;
         }
 
+        public Task<MetaMorpheusEngineResults> RunAsync() => Task.Run(Run);
+
         /// <summary>
         /// Changes the name of the analytes from "peptide" to "proteoform" or "oligo" if the protease is set to top-down
         /// </summary>

MetaMorpheus/EngineLayer/ProteinScoringAndFdr/ProteinScoringAndFdrEngine.cs

@@ -12,6 +12,7 @@ public class ProteinScoringAndFdrEngine : MetaMorpheusEngine
         private readonly bool TreatModPeptidesAsDifferentPeptides;
         private readonly bool MergeIndistinguishableProteinGroups;
         private readonly List<ProteinGroup> ProteinGroups;
+        private readonly HashSet<string> _decoyIdentifiers;
 
         public ProteinScoringAndFdrEngine(List<ProteinGroup> proteinGroups, List<SpectralMatch> newPsms, bool noOneHitWonders, bool treatModPeptidesAsDifferentPeptides, bool mergeIndistinguishableProteinGroups, CommonParameters commonParameters, List<(string fileName, CommonParameters fileSpecificParameters)> fileSpecificParameters, List<string> nestedIds) : base(commonParameters, fileSpecificParameters, nestedIds)
         {
@@ -20,6 +21,7 @@ public ProteinScoringAndFdrEngine(List<ProteinGroup> proteinGroups, List<Spectra
             NoOneHitWonders = noOneHitWonders;
             TreatModPeptidesAsDifferentPeptides = treatModPeptidesAsDifferentPeptides;
             MergeIndistinguishableProteinGroups = mergeIndistinguishableProteinGroups;
+            _decoyIdentifiers = proteinGroups.SelectMany(p => p.Proteins.Where(b => b.IsDecoy).Select(b => b.Accession.Split('_')[0])).ToHashSet();
         }
 
         protected override MetaMorpheusEngineResults RunSpecific()
@@ -31,9 +33,12 @@ protected override MetaMorpheusEngineResults RunSpecific()
             return myAnalysisResults;
         }
 
-        private static string StripDecoyIdentifier(string proteinGroupName) //we're keeping only the better scoring protein group for each target/decoy pair. to do that we need to strip decoy from the name temporarily. this is the "top-picked" method
+        private static string StripDecoyIdentifier(string proteinGroupName, HashSet<string> decoyIdentifiers) //we're keeping only the better scoring protein group for each target/decoy pair. to do that we need to strip decoy from the name temporarily. this is the "top-picked" method
         {
-            return proteinGroupName.Contains("DECOY_") ? proteinGroupName.Replace("DECOY_", "") : proteinGroupName;
+            foreach (var ident in decoyIdentifiers.Where(proteinGroupName.Contains))
+                return proteinGroupName.Replace($"{ident}_", "");
+
+            return proteinGroupName;
         }
 
         private void ScoreProteinGroups(List<ProteinGroup> proteinGroups, IEnumerable<SpectralMatch> psmList)
@@ -142,7 +147,7 @@ private List<ProteinGroup> DoProteinFdr(List<ProteinGroup> proteinGroups)
             {
                 foreach (var protein in pg.Proteins)
                 {
-                    string stippedAccession = StripDecoyIdentifier(protein.Accession); //remove "DECOY_" from the accession
+                    string stippedAccession = StripDecoyIdentifier(protein.Accession, _decoyIdentifiers); //remove "DECOY_" from the accession
 
                     if (accessionToProteinGroup.TryGetValue(stippedAccession, out List<ProteinGroup> groups))
                     {

MetaMorpheus/GUI/ForDisplayingInDataGrids/ProteinDbForDataGrid.cs

@@ -1,9 +1,12 @@
 using System.IO;
+using EngineLayer;
+using EngineLayer.DatabaseLoading;
+using GuiFunctions;
 using TaskLayer;
 
 namespace MetaMorpheusGUI
 {
-    public class ProteinDbForDataGrid
+    public class ProteinDbForDataGrid : BaseViewModel
     {
         #region Public Constructors
 
@@ -27,17 +30,41 @@ public ProteinDbForDataGrid(DbForTask uu)
             Contaminant = uu.IsContaminant;
             FilePath = uu.FilePath;
             FileName = uu.FileName;
+            DecoyIdentifier = uu.DecoyIdentifier;
         }
 
         #endregion Public Constructors
 
         #region Public Properties
 
-        public bool Use { get; set; }
-        public bool Contaminant { get; set; }
+        private bool _use;
+        private bool _isContaminant;
+        private bool _inProgress;
+        private string _decoyIdentifier = GlobalVariables.DecoyIdentifier;
+
+        public bool Use
+        {
+            get => _use;
+            set { _use = value; OnPropertyChanged(nameof(Use)); }
+        }
+        public bool Contaminant
+        {
+            get => _isContaminant;
+            set { _isContaminant = value; OnPropertyChanged(nameof(Contaminant)); }
+        }
         public string FileName { get; private set; }
         public string FilePath { get; private set; }
-        public bool InProgress { get; private set; }
+        public bool InProgress
+        {
+            get => _inProgress;
+            private set { _inProgress = value; OnPropertyChanged(nameof(InProgress)); }
+        }
+
+        public string DecoyIdentifier
+        {
+            get => _decoyIdentifier;
+            set { _decoyIdentifier = value; OnPropertyChanged(nameof(DecoyIdentifier)); }
+        }
 
         #endregion Public Properties
 

MetaMorpheus/GUI/GUI.csproj

@@ -62,7 +62,7 @@
     <PackageReference Include="Microsoft.ML.CpuMath" Version="3.0.1" />
     <PackageReference Include="Microsoft.ML.FastTree" Version="3.0.1" />
     <PackageReference Include="Microsoft.NETCore.App" Version="2.2.8" />
-    <PackageReference Include="mzLib" Version="1.0.569" />
+    <PackageReference Include="mzLib" Version="5.3.1" />
     <PackageReference Include="Nett" Version="0.15.0" />
     <PackageReference Include="Newtonsoft.Json" Version="13.0.3" />
     <PackageReference Include="OxyPlot.Core" Version="2.0.0" />

MetaMorpheus/GUI/MainWindow.xaml

@@ -435,10 +435,11 @@
                             <Style TargetType="{x:Type DataGridCell}" BasedOn="{StaticResource DataGridCellStyle}" />
                         </DataGrid.CellStyle>
                         <DataGrid.Columns>
-                            <DataGridTextColumn Header="File" Binding="{Binding FileName, Mode=OneWay}" Width="250" />
-                            <DataGridTextColumn Header="File Path" Binding="{Binding FilePath, Mode=OneWay}" Width="200" />
-                            <DataGridTextColumn Header="Contaminant?" Binding="{Binding Contaminant, Mode=OneWay}" Width="250" />
-                            <DataGridCheckBoxColumn Header="Use?" Binding="{Binding Use, Mode=TwoWay}"/>
+                            <DataGridTextColumn Header="File" Binding="{Binding FileName, Mode=OneWay}" Width="366" />
+                            <DataGridTextColumn Header="File Path" Binding="{Binding FilePath, Mode=OneWay}" Width="160" />
+                            <DataGridCheckBoxColumn Header="Contaminant?" Binding="{Binding Contaminant, Mode=TwoWay}" MinWidth="90" />
+                            <DataGridCheckBoxColumn Header="Use?" Binding="{Binding Use, Mode=TwoWay}" MinWidth="80"/>
+                            <DataGridTextColumn Header="Decoy Identifier" Binding="{Binding DecoyIdentifier, Mode=TwoWay}" Width="160"/>
                         </DataGrid.Columns>
                     </DataGrid>
                 </Grid>
@@ -583,8 +584,9 @@
                                 </DataGrid.CellStyle>
                                 <DataGrid.Columns>
                                     <DataGridTextColumn Header="File" Binding="{Binding FileName, Mode=OneWay}" Width="230" />
-                                    <DataGridTextColumn Header="Contaminant?" Binding="{Binding Contaminant, Mode=OneWay}" Width="170" />
-                                    <DataGridCheckBoxColumn Header="Use?" Binding="{Binding Use, Mode=TwoWay}"/>
+                                    <DataGridCheckBoxColumn Header="Use?" Binding="{Binding Use, Mode=TwoWay}" MinWidth="40"/>
+                                    <DataGridCheckBoxColumn Header="Contaminant?" Binding="{Binding Contaminant, Mode=TwoWay}" MinWidth="90"/>
+                                    <DataGridTextColumn Header="Decoy Identifier" Binding="{Binding DecoyIdentifier, Mode=TwoWay}" Width="160" />
                                 </DataGrid.Columns>
                             </DataGrid>
                             <Button Name="MiniAddProteinDbButton" Grid.Row="0" Width="20" Height="20" HorizontalAlignment="Right" VerticalAlignment="Bottom" Margin="0,0,1,1" Content="+" 

MetaMorpheus/GUI/MainWindow.xaml.cs

@@ -22,6 +22,7 @@
 using System.Text.RegularExpressions;
 using Readers.InternalResults;
 using System.Diagnostics;
+using EngineLayer.DatabaseLoading;
 
 namespace MetaMorpheusGUI
 {
@@ -668,7 +669,8 @@ private void DatabaseOrSpectraFile_DoubleClick(object sender, MouseButtonEventAr
             }
 
             // user is probably just checking or unchecking a checkbox, don't open the file
-            if (sender is DataGridCell cell && cell.Column is DataGridCheckBoxColumn)
+            // User is double clicking the decoy ident column. 
+            if (sender is DataGridCell { Column: DataGridCheckBoxColumn } || sender is DataGridCell { Column: DataGridTextColumn, TabIndex: >= 3})
             {
                 return;
             }
@@ -989,7 +991,7 @@ private void RunAllTasks_Click(object sender, RoutedEventArgs e)
             // everything is ready to run
             EverythingRunnerEngine a = new EverythingRunnerEngine(InProgressTasks.Select(b => (b.DisplayName, b.Task)).ToList(),
                 SpectraFiles.Where(b => b.Use).Select(b => b.FilePath).ToList(),
-                ProteinDatabases.Where(b => b.Use).Select(b => new DbForTask(b.FilePath, b.Contaminant)).ToList(),
+                ProteinDatabases.Where(b => b.Use).Select(b => new DbForTask(b.FilePath, b.Contaminant, b.DecoyIdentifier)).ToList(),
                 outputFolder);
 
             var t = new Task(a.Run);
@@ -1091,11 +1093,12 @@ private void BoxWithList_PreviewKeyDown(object sender, KeyEventArgs e)
         {
             if (!RunTasksButton.IsEnabled) return;
 
+
             switch (e.Key)
             {
                 // delete selected task/db/spectra
-                case Key.Delete:
-                case Key.Back:
+                case Key.Delete when e.OriginalSource is DataGrid:
+                case Key.Back when e.OriginalSource is DataGrid:
                     Delete_Click(sender, e);
                     e.Handled = true;
                     break;

MetaMorpheus/GUI/Util/GuiGlobalParams.cs

@@ -11,6 +11,7 @@ public class GuiGlobalParams
 
         // User can set a custom proteome directory. Be sure to use double slashes in the path, otherwise it will not be read in properly. 
         public string UserSpecifiedProteomeDir { get; internal set; } = "";
+        public string DecoyIdentifier { get; internal set; } = "DECOY";
 
         //Ask about protease-specific parameter recommendations
         public bool AskAboutTopDownParams { get; internal set; } = true;

MetaMorpheus/GUI/Util/UpdateGUISettings.cs

@@ -23,6 +23,7 @@ public static bool LoadGUISettings()
                 Toml.WriteFile(Params, Path.Combine(GlobalVariables.DataDir, @"GUIsettings.toml"), MetaMorpheusTask.tomlConfig);
             }
 
+            GlobalVariables.DecoyIdentifier = Params.DecoyIdentifier ??= "DECOY";
             if (GlobalVariables.MetaMorpheusVersion.Contains("Not a release version"))
             {
                 Params.AskAboutUpdating = false;

MetaMorpheus/GuiFunctions/GuiFunctions.csproj

@@ -16,7 +16,7 @@
   <ItemGroup>
     <PackageReference Include="itext7" Version="8.0.5" />
     <PackageReference Include="itext7.bouncy-castle-adapter" Version="8.0.5" />
-    <PackageReference Include="mzLib" Version="1.0.569" />
+    <PackageReference Include="mzLib" Version="5.3.1" />
     <PackageReference Include="OxyPlot.Wpf" Version="2.0.0" />
     <PackageReference Include="Svg" Version="3.4.7" />
   </ItemGroup>

MetaMorpheus/GuiFunctions/MetaDraw/BioPolymerCoverage/BioPolymerTabViewModel.cs

@@ -13,6 +13,7 @@
 using System.Windows.Media.Imaging;
 using Easy.Common.Extensions;
 using EngineLayer;
+using EngineLayer.DatabaseLoading;
 using MzLibUtil;
 using Omics;
 using Readers;
@@ -95,13 +96,10 @@ private async void LoadDatabase()
         try
         {
             // Load biopolymers asynchronously
-            var bioPolymers = await Task.Run(() =>
-                new SearchTask().LoadBioPolymers(
-                    "", new() { new DbForTask(DatabasePath, false) }, true, DecoyType.None, new(), new())
-                .ToDictionary(p => p.Accession, p => p)
-            );
+            var dbLoader = new DatabaseLoadingEngine(new(), [], [], [new DbForTask(DatabasePath, false)], "", DecoyType.None);
+            var loadingResults = await dbLoader.RunAsync();
 
-            _allBioPolymers = bioPolymers;
+            _allBioPolymers = (loadingResults as DatabaseLoadingEngineResults)!.BioPolymers.ToDictionary(bp => bp.Accession, bp => bp);
 
             if (_allBioPolymers.Count == 0)
             {

MetaMorpheus/TaskLayer/AveragingTask/SpectralAveragingTask.cs

@@ -5,6 +5,7 @@
 using System.Text;
 using System.Threading.Tasks;
 using EngineLayer;
+using EngineLayer.DatabaseLoading;
 using FlashLFQ;
 using IO.MzML;
 using MassSpectrometry;