RNA in the GUI

MetaMorpheus/EngineLayer/DatabaseLoading/DatabaseLoadingEngine.cs

@@ -23,7 +23,7 @@
     string taskId,
     DecoyType decoyType,
     bool generateTargets = true,
     List<string> localizableMods = null,
     TargetContaminantAmbiguity tcAmbiguity = TargetContaminantAmbiguity.RemoveContaminant)
     : MetaMorpheusEngine(commonParameters, fileSpecificParameters, nestedIds)
 {
@@ -120,11 +120,12 @@
 
         if (theExtension.Equals(".fasta") || theExtension.Equals(".fa"))
         {
             unknownMods = null;
             rnaList = RnaDbLoader.LoadRnaFasta(fileName, generateTargets, decoyType, isContaminant, out var dbErrors);
         }
         else
         {
+            GlobalVariables.AddMods(ProteinDbLoader.GetPtmListFromProteinXml(fileName), true, true);
             // TODO: Add in variant params when fixed in MzLib. 
             List<string> modTypesToExclude = GlobalVariables.AllRnaModTypesKnown.Where(b => !localizeableModificationTypes.Contains(b)).ToList();
             rnaList = RnaDbLoader.LoadRnaXML(fileName, generateTargets, decoyType, isContaminant, GlobalVariables.AllRnaModsKnown, modTypesToExclude, out unknownMods, commonParameters.MaxThreadsToUsePerFile, decoyIdentifier: decoyIdentifier);
@@ -146,7 +147,7 @@
 
         if (theExtension.Equals(".fasta") || theExtension.Equals(".fa"))
         {
             um = null;
             proteinList = ProteinDbLoader.LoadProteinFasta(fileName, generateTargets, decoyType, isContaminant, out var dbErrors,
                 ProteinDbLoader.UniprotAccessionRegex, ProteinDbLoader.UniprotFullNameRegex, ProteinDbLoader.UniprotFullNameRegex, ProteinDbLoader.UniprotGeneNameRegex,
                 ProteinDbLoader.UniprotOrganismRegex, commonParameters.MaxThreadsToUsePerFile, addTruncations: commonParameters.AddTruncations);
@@ -154,10 +155,10 @@
         else
         {
             List<string> modTypesToExclude = GlobalVariables.AllModTypesKnown.Where(b => !localizeableModificationTypes.Contains(b)).ToList();
+            GlobalVariables.AddMods(ProteinDbLoader.GetPtmListFromProteinXml(fileName), true, false);
             //proteinList = ProteinDbLoader.LoadProteinXML(fileName, generateTargets, decoyType, GlobalVariables.AllModsKnown, isContaminant, modTypesToExclude, out um, commonParameters.MaxThreadsToUsePerFile, commonParameters.MaxHeterozygousVariants, commonParameters.MinVariantDepth, addTruncations: commonParameters.AddTruncations, decoyIdentifier: decoyIdentifier);
             proteinList = ProteinDbLoader.LoadProteinXML(fileName, generateTargets, decoyType, GlobalVariables.AllModsKnown, isContaminant, modTypesToExclude, out um, commonParameters.MaxThreadsToUsePerFile, 0, commonParameters.MinVariantDepth, addTruncations: commonParameters.AddTruncations, decoyIdentifier: decoyIdentifier);
         }
-
         emptyEntriesCount = proteinList.Count(p => p.BaseSequence.Length == 0);
         return proteinList.Where(p => p.BaseSequence.Length > 0);
     }
@@ -286,7 +287,7 @@
                 continue;
 
             // Use a pooled list to avoid repeated allocations in tight loops
             List<string> peptidesToReplace = null;
             foreach (var peptide in bioPolymer.Digest(digestionParams, variable, fixedM))
             {
                 if (targetPeptideSequences.Contains(peptide.BaseSequence))

MetaMorpheus/EngineLayer/FdrAnalysis/FdrAnalysisEngine.cs

@@ -1,8 +1,10 @@
 using EngineLayer.CrosslinkSearch;
 using EngineLayer.SpectrumMatch;
+using Proteomics.ProteolyticDigestion;
 using System;
 using System.Collections.Generic;
 using System.Linq;
+using Transcriptomics.Digestion;
 
 namespace EngineLayer.FdrAnalysis
 {
@@ -386,25 +388,19 @@ public static void PepQValueInverted(List<SpectralMatch> psms, bool peptideLevel
 
         public static void Compute_PEPValue(FdrAnalysisResults myAnalysisResults, List<SpectralMatch> psms, List<(string fileName, CommonParameters fileSpecificParameters)> fileSpecificParameters, string outputFolder)
         {
-            string searchType;
             // Currently, searches of mixed data (bottom-up + top-down) are not supported
             // PEP will be calculated based on the search type of the first file/PSM in the list, which isn't ideal
             // This will be addressed in a future release
-            switch(psms[0].DigestionParams.DigestionAgent.Name)
-            {
-               case "top-down":
-                    searchType = "top-down";
-                    break;
-                default:
-                    searchType = "standard";
-                    break;
-            }
-            if (psms[0] is CrosslinkSpectralMatch)
+
+            string searchType = psms[0] switch
             {
-                searchType = "crosslink";
-            }
-            myAnalysisResults.BinarySearchTreeMetrics = new PepAnalysisEngine(psms, searchType, fileSpecificParameters, outputFolder).ComputePEPValuesForAllPSMs();
+                OligoSpectralMatch => "RNA",
+                PeptideSpectralMatch psm when psm.DigestionParams.DigestionAgent is Protease { Name: "top-down" } => "top-down",
+                CrosslinkSpectralMatch => "crosslink",
+                _ => "standard"
+            };
 
+            myAnalysisResults.BinarySearchTreeMetrics = new PepAnalysisEngine(psms, searchType, fileSpecificParameters, outputFolder).ComputePEPValuesForAllPSMs();
         }
 
         /// <summary>

MetaMorpheus/EngineLayer/FdrAnalysis/PEPAnalysisEngine.cs

@@ -367,20 +367,20 @@ public static string AggregateMetricsForOutput(List<CalibratedBinaryClassificati
             s.AppendLine("************************************************************");
             s.AppendLine("*       Metrics for Determination of PEP Using Binary Classification      ");
             s.AppendLine("*-----------------------------------------------------------");
-            s.AppendLine("*       Accuracy:  " + accuracy.Average().ToString());
-            s.AppendLine("*       Area Under Curve:  " + areaUnderRocCurve.Average().ToString());
-            s.AppendLine("*       Area under Precision recall Curve:  " + areaUnderPrecisionRecallCurve.Average().ToString());
-            s.AppendLine("*       F1Score:  " + F1Score.Average().ToString());
-            s.AppendLine("*       LogLoss:  " + logLossAverage.ToString());
-            s.AppendLine("*       LogLossReduction:  " + logLossReductionAverage.ToString());
-            s.AppendLine("*       PositivePrecision:  " + positivePrecision.Average().ToString());
-            s.AppendLine("*       PositiveRecall:  " + positiveRecall.Average().ToString());
-            s.AppendLine("*       NegativePrecision:  " + negativePrecision.Average().ToString());
-            s.AppendLine("*       NegativeRecall:  " + negativeRecall.Average().ToString());
-            s.AppendLine("*       Count of Ambiguous Peptides Removed:  " + sumOfAllAmbiguousPeptidesResolved.ToString());
+            s.AppendLine("*       Accuracy:  " + accuracy.Average());
+            s.AppendLine("*       Area Under Curve:  " + areaUnderRocCurve.Average());
+            s.AppendLine("*       Area under Precision recall Curve:  " + areaUnderPrecisionRecallCurve.Average());
+            s.AppendLine("*       F1Score:  " + F1Score.Average());
+            s.AppendLine("*       LogLoss:  " + logLossAverage);
+            s.AppendLine("*       LogLossReduction:  " + logLossReductionAverage);
+            s.AppendLine("*       PositivePrecision:  " + positivePrecision.Average());
+            s.AppendLine("*       PositiveRecall:  " + positiveRecall.Average());
+            s.AppendLine("*       NegativePrecision:  " + negativePrecision.Average());
+            s.AppendLine("*       NegativeRecall:  " + negativeRecall.Average());
+            s.AppendLine($"*       Count of Ambiguous {char.ToUpper(GlobalVariables.AnalyteType.GetUniqueFormLabel()[0]) + GlobalVariables.AnalyteType.GetUniqueFormLabel()[1..]}s Removed:  " + sumOfAllAmbiguousPeptidesResolved);
             s.AppendLine("*       Q-Value Cutoff for Training Targets:  " + qValueCutoff);
-            s.AppendLine("*       Targets Used for Training:  " + positiveTrainingCount.ToString());
-            s.AppendLine("*       Decoys Used for Training:  " + negativeTrainingCount.ToString());
+            s.AppendLine("*       Targets Used for Training:  " + positiveTrainingCount);
+            s.AppendLine("*       Decoys Used for Training:  " + negativeTrainingCount);
             s.AppendLine("************************************************************");
             return s.ToString();
         }
@@ -539,23 +539,24 @@ public PsmData CreateOnePsmDataEntry(string searchType, SpectralMatch psm, Spect
                 if (psm.DigestionParams.DigestionAgent.Name != "top-down")
                 {
                     missedCleavages = tentativeSpectralMatch.SpecificBioPolymer.MissedCleavages;
-                    bool fileIsCzeSeparationType = FileSpecificParametersDictionary.ContainsKey(Path.GetFileName(psm.FullFilePath)) && FileSpecificParametersDictionary[Path.GetFileName(psm.FullFilePath)].SeparationType == "CZE";
-
-                    if (!fileIsCzeSeparationType)
+                    var fileName = Path.GetFileName(psm.FullFilePath);
+                    bool fileIsCzeSeparationType = FileSpecificParametersDictionary.TryGetValue(fileName, out var fileParams) && fileParams.SeparationType == "CZE";
+
+                    if (searchType != "RNA")
                     {
-                        if (tentativeSpectralMatch.SpecificBioPolymer.BaseSequence.Equals(tentativeSpectralMatch.SpecificBioPolymer.FullSequence))
+                        if (!fileIsCzeSeparationType)
                         {
-                            hydrophobicityZscore = (float)Math.Round(GetSSRCalcHydrophobicityZScore(psm, tentativeSpectralMatch.SpecificBioPolymer, FileSpecificTimeDependantHydrophobicityAverageAndDeviation_unmodified) * 10.0, 0);
+                            var isUnmodified = tentativeSpectralMatch.SpecificBioPolymer.BaseSequence.Equals(tentativeSpectralMatch.SpecificBioPolymer.FullSequence);
+                            var dict = isUnmodified
+                                ? FileSpecificTimeDependantHydrophobicityAverageAndDeviation_unmodified
+                                : FileSpecificTimeDependantHydrophobicityAverageAndDeviation_modified;
+                            hydrophobicityZscore = (float)Math.Round(GetSSRCalcHydrophobicityZScore(psm, tentativeSpectralMatch.SpecificBioPolymer, dict) * 10.0, 0);
                         }
                         else
                         {
-                            hydrophobicityZscore = (float)Math.Round(GetSSRCalcHydrophobicityZScore(psm, tentativeSpectralMatch.SpecificBioPolymer, FileSpecificTimeDependantHydrophobicityAverageAndDeviation_modified) * 10.0, 0);
+                            hydrophobicityZscore = (float)Math.Round(GetMobilityZScore(psm, tentativeSpectralMatch.SpecificBioPolymer) * 10.0, 0);
                         }
                     }
-                    else
-                    {
-                        hydrophobicityZscore = (float)Math.Round(GetMobilityZScore(psm, tentativeSpectralMatch.SpecificBioPolymer) * 10.0, 0);
-                    }
                 }
                 //this is not for actual crosslinks but for the byproducts of crosslink loop links, deadends, etc.
                 if (psm is CrosslinkSpectralMatch)

MetaMorpheus/EngineLayer/FdrAnalysis/PsmData.cs

@@ -40,6 +40,14 @@ public class PsmData
                     "PrecursorChargeDiffToMode", "DeltaScore", "AlphaIntensity", "BetaIntensity",
                     "LongestFragmentIonSeries_Alpha", "LongestFragmentIonSeries_Beta", "IsInter", "IsIntra"
                 }
+            },
+            {
+                "RNA",
+                new []
+                { "TotalMatchingFragmentCount", "Intensity", "PrecursorChargeDiffToMode", "DeltaScore", 
+                    "Notch", "ModsCount", "AbsoluteAverageFragmentMassErrorFromMedian","MissedCleavagesCount",
+                    "Ambiguity", "LongestFragmentIonSeries", "ComplementaryIonCount"
+                }
             }
         }.ToImmutableDictionary();
 

MetaMorpheus/EngineLayer/GlobalVariables.cs

@@ -11,6 +11,7 @@
 using System.Diagnostics;
 using System.IO;
 using System.Linq;
+using System.Reflection;
 using System.Runtime.InteropServices;
 using System.Text;
 using TopDownProteomics;
@@ -89,22 +90,25 @@ public static void SetUpGlobalVariables()
             LoadAvailableProteomes();
         }
 
-        public static void AddMods(IEnumerable<Modification> modifications, bool modsAreFromTheTopOfProteinXml)
+        public static void AddMods(IEnumerable<Modification> modifications, bool modsAreFromTheTopOfProteinXml, bool isRna = false)
         {
+            var allMods = isRna ? _AllRnaModsKnown : _AllModsKnown;
+            var modTypes = isRna ? _AllRnaModTypesKnown : _AllModTypesKnown;
+
             foreach (var mod in modifications)
             {
                 if (string.IsNullOrEmpty(mod.ModificationType) || string.IsNullOrEmpty(mod.IdWithMotif))
                 {
                     ErrorsReadingMods.Add(mod.ToString() + Environment.NewLine + " has null or empty modification type");
                     continue;
                 }
-                if (AllModsKnown.Any(b => b.IdWithMotif.Equals(mod.IdWithMotif) && b.ModificationType.Equals(mod.ModificationType) && !b.Equals(mod)))
+                if (allMods.Any(b => b.IdWithMotif.Equals(mod.IdWithMotif) && b.ModificationType.Equals(mod.ModificationType) && !b.Equals(mod)))
                 {
                     if (modsAreFromTheTopOfProteinXml)
                     {
-                        _AllModsKnown.RemoveAll(p => p.IdWithMotif.Equals(mod.IdWithMotif) && p.ModificationType.Equals(mod.ModificationType) && !p.Equals(mod));
-                        _AllModsKnown.Add(mod);
-                        _AllModTypesKnown.Add(mod.ModificationType);
+                        allMods.RemoveAll(p => p.IdWithMotif.Equals(mod.IdWithMotif) && p.ModificationType.Equals(mod.ModificationType) && !p.Equals(mod));
+                        allMods.Add(mod);
+                        modTypes.Add(mod.ModificationType);
                     }
                     else
                     {
@@ -113,36 +117,34 @@ public static void AddMods(IEnumerable<Modification> modifications, bool modsAre
                     }
                     continue;
                 }
-                else if (AllModsKnown.Any(b => b.IdWithMotif.Equals(mod.IdWithMotif) && b.ModificationType.Equals(mod.ModificationType)))
+                if (allMods.Any(b => b.IdWithMotif.Equals(mod.IdWithMotif) && b.ModificationType.Equals(mod.ModificationType)))
                 {
                     // same ID, same mod type, and same mod properties; continue and don't output an error message
                     // this could result from reading in an XML database with mods annotated at the top
                     // that are already loaded in MetaMorpheus
                     continue;
                 }
-                else if (AllModsKnown.Any(m => m.IdWithMotif == mod.IdWithMotif))
+                if (allMods.Any(m => m.IdWithMotif == mod.IdWithMotif))
                 {
                     // same ID but different mod types. This can happen if the user names a mod the same as a UniProt mod
                     // this is problematic because if a mod is annotated in the database, all we have to go on is an ID ("description" tag).
                     // so we don't know which mod to use, causing unnecessary ambiguity
                     if (modsAreFromTheTopOfProteinXml)
                     {
-                        _AllModsKnown.RemoveAll(p => p.IdWithMotif.Equals(mod.IdWithMotif) && !p.Equals(mod));
-                        _AllModsKnown.Add(mod);
-                        _AllModTypesKnown.Add(mod.ModificationType);
+                        allMods.RemoveAll(p => p.IdWithMotif.Equals(mod.IdWithMotif) && !p.Equals(mod));
+                        allMods.Add(mod);
+                        modTypes.Add(mod.ModificationType);
                     }
                     else if (!mod.ModificationType.Equals("Unimod"))
                     {
                         ErrorsReadingMods.Add("Duplicate mod IDs! Skipping " + mod.ModificationType + ":" + mod.IdWithMotif);
                     }
                     continue;
                 }
-                else
-                {
-                    // no errors! add the mod
-                    _AllModsKnown.Add(mod);
-                    _AllModTypesKnown.Add(mod.ModificationType);
-                }
+
+                // no errors! add the mod
+                allMods.Add(mod);
+                modTypes.Add(mod.ModificationType);
             }
         }
 
@@ -408,6 +410,8 @@ private static void LoadModifications()
                     // Glycan modifications are handled separately in LoadGlycans()
                     continue;
                 }
+                if (modFile.Contains("Rna"))
+                    continue;
                 AddMods(PtmListLoader.ReadModsFromFile(modFile, out var errorMods), false);
             }
 
@@ -441,10 +445,14 @@ private static void LoadRnaModifications()
             using (var reader = new StreamReader(stream))
             {
                 string fileContent = reader.ReadToEnd();
-                foreach (var mod in PtmListLoader.ReadModsFromString(fileContent, out var errors))
-                {
-                    _AllRnaModsKnown.Add(mod);
-                }
+                var mods = PtmListLoader.ReadModsFromString(fileContent, out var errors);
+                AddMods(mods, false, true);
+            }
+
+            var customModsPath = Path.Combine(DataDir, @"Mods", "RnaCustomModifications.txt");
+            if (File.Exists(customModsPath))
+            {
+                AddMods(PtmListLoader.ReadModsFromFile(customModsPath, out var errorMods), false, true);
             }
 
             // populate mod types and dictionary

MetaMorpheus/EngineLayer/MetaMorpheusEngine.cs

@@ -149,6 +149,7 @@ public static List<MatchedFragmentIon> MatchFragmentIons(Ms2ScanWithSpecificMass
                         continue;
                     }
 
+                    // Magic number represents mzbinning space. 
                     double theoreticalFragmentMz = Math.Round(product.NeutralMass.ToMz(1) / 1.0005079, 0) * 1.0005079;
                     var closestMzIndex = scan.TheScan.MassSpectrum.GetClosestPeakIndex(theoreticalFragmentMz);
 

MetaMorpheus/EngineLayer/Mods/RnaMods.txt

@@ -28,13 +28,6 @@ MT   Biological
 CF   C2 H4
 DR   Unimod; 34.
 //
-ID   DeoxyFluoronation
-TG   A or C or G or U
-PP   Anywhere.
-MT   Biological
-CF   O-1 H-1 F1
-DR   Unimod; 34.
-//
 ID   MethoxyEthoxylation
 TG   A or C or G or U
 PP   Anywhere.
@@ -57,12 +50,12 @@ CF   N-1H-1O-1
 DR   Unimod; 34.
 //
 ################################## Common Artificial
-ID   2' Fluronation 
-TG   X
+ID   DeoxyFluoronation
+TG   A or C or G or U
 PP   Anywhere.
-MT   Artificial
-CF   H-1 O-1 F1
-DR   Unimod; 280.
+MT   Common Artificial
+CF   O-1 H-1 F1
+DR   Unimod; 34.
 //
 ################################## Terminal Shifts
 ID   Cyclic Phosphate

MetaMorpheus/GUI/App.xaml

@@ -8,6 +8,7 @@
         <!--TODO: store these in a file and read them from disk, to let users style MetaMorpheus?-->
         <SolidColorBrush x:Key="SideBarMainColor" Color="#353b48" />
         <SolidColorBrush x:Key="AccentColor" Color="#0097e6" />
+        <SolidColorBrush x:Key="LightAccentColor" Color="#bdf0ff" />
         <SolidColorBrush x:Key="PurpleAccentColor" Color="#e6aeff" />
         <SolidColorBrush x:Key="SecondaryButtonColor" Color="#504B59" />
         <SolidColorBrush x:Key="BackgroundColor" Color="#f5f6fa" />
@@ -20,6 +21,10 @@
         <Color x:Key="ForegroundDark">#bdbdbd</Color>
         <SolidColorBrush x:Key="ForegroundDarkBrush" Color="{StaticResource ForegroundDark}" />
 
+        <SolidColorBrush x:Key="RnaAccentColor" Color="#a64ef9" />
+        <SolidColorBrush x:Key="LightRnaAccentColor" Color="#FFC4C4FF" />
+        <SolidColorBrush x:Key="RnaSideBarMainColor" Color="#433d49" />
+
         <!--Default button style for all of MetaMorpheus-->
         <!--<Style TargetType="Button">
             <Setter Property="BorderThickness" Value="0"/>
@@ -100,5 +105,13 @@
                 </Setter.Value>
             </Setter>
         </Style>
+
+        <!-- Icons -->
+        <BitmapImage x:Key="ProteinIcon" UriSource="Icons/IconProtein.bmp"/>
+        <BitmapImage x:Key="RnaIcon" UriSource="Icons/IconRna.bmp"/>
+
+        <!-- Converters -->
+        <local:ProteinRnaTerminologyConverter x:Key="ProteinRnaTerminologyConverter"/>
+        <local:CollapseOnRnaModeConverter x:Key="CollapseOnRnaModeConverter"/>
     </Application.Resources>
 </Application>

MetaMorpheus/GUI/GUI.csproj

@@ -43,6 +43,7 @@
     <None Remove="Icons\mail_20.bmp" />
     <None Remove="Icons\MMnice.ico" />
     <None Remove="Icons\proxl-ms.bmp" />
+    <None Remove="Icons\rna_4.bmp" />
     <None Remove="Icons\Slack_Mark_Web.bmp" />
     <None Remove="Icons\spritz20.png" />
     <None Remove="Icons\Twitter_Logo_Blue.bmp" />
@@ -107,9 +108,15 @@
     <Resource Include="Icons\MMnice.ico">
       <CopyToOutputDirectory>Always</CopyToOutputDirectory>
     </Resource>
+    <Resource Include="Icons\IconProtein.bmp">
+      <CopyToOutputDirectory>Always</CopyToOutputDirectory>
+    </Resource>
     <Resource Include="Icons\proxl-ms.bmp">
       <CopyToOutputDirectory>Always</CopyToOutputDirectory>
     </Resource>
+    <Resource Include="Icons\IconRna.bmp">
+      <CopyToOutputDirectory>Always</CopyToOutputDirectory>
+    </Resource>
     <Resource Include="Icons\Slack_Mark_Web.bmp">
       <CopyToOutputDirectory>Always</CopyToOutputDirectory>
     </Resource>