Update Omics to read additional SpectrumMatchFromTsv types

mzLib/Omics/SpectrumMatch/CrosslinkLibrarySpectrum.cs

@@ -0,0 +1,124 @@
+using Omics.Fragmentation;
+using System;
+using System.Collections.Generic;
+using System.Linq;
+using System.Text;
+using System.Text.RegularExpressions;
+using System.Threading.Tasks;
+
+namespace Omics.SpectrumMatch
+{
+        public class CrosslinkLibrarySpectrum : LibrarySpectrum
+        {
+            public CrosslinkLibrarySpectrum BetaPeptideSpectrum { get; }
+            public string AlphaPeptideSequence { get; private set; }
+            public string BetaPeptideSequence { get; private set; }
+            public string UniqueSequence { get; private set; }
+            public bool IsBetaPeptide { get; }
+            public static Regex CrosslinkRegex = new Regex(@"\(\d+\)");
+            public new string Name => UniqueSequence + "/" + ChargeState;
+
+            public CrosslinkLibrarySpectrum(
+                string uniqueSequence,
+                double precursorMz,
+                int precursorCharge,
+                List<MatchedFragmentIon> peaks,
+                double rt,
+                List<MatchedFragmentIon> betaPeaks,
+                bool isDecoy = false) : this(
+                    uniqueSequence,
+                    precursorMz,
+                    precursorCharge,
+                    peaks,
+                    rt,
+                    new CrosslinkLibrarySpectrum(uniqueSequence, precursorMz, precursorCharge, betaPeaks, rt),
+                    isDecoy)
+            { }
+
+            public CrosslinkLibrarySpectrum(
+                string uniqueSequence,
+                double precursorMz,
+                int precursorCharge,
+                List<MatchedFragmentIon> peaks,
+                double rt,
+                CrosslinkLibrarySpectrum betaSpectrum = null,
+                bool isDecoy = false) : base(uniqueSequence, precursorMz, precursorCharge, peaks, rt, isDecoy)
+            {
+                UniqueSequence = uniqueSequence;
+                if (betaSpectrum == null)
+                {
+                    IsBetaPeptide = true;
+                }
+                else
+                {
+                    BetaPeptideSpectrum = betaSpectrum;
+                }
+                SetAlphaBetaSequence();
+            }
+
+            private void SetAlphaBetaSequence()
+            {
+                string[] uniqueSequenceSplit = CrosslinkRegex.Split(UniqueSequence);
+                if (uniqueSequenceSplit.Length >= 2)
+                {
+                    AlphaPeptideSequence = uniqueSequenceSplit[0];
+                    BetaPeptideSequence = uniqueSequenceSplit[1];
+                }
+                else
+                {
+                    AlphaPeptideSequence = null;
+                    BetaPeptideSequence = null;
+                }
+            }
+
+            public override string ToString()
+            {
+                StringBuilder spectrum = new();
+                spectrum.AppendLine("Name: " + Name);
+                spectrum.AppendLine("MW: " + PrecursorMz);
+                spectrum.Append("Comment: ");
+                spectrum.Append("Parent=" + PrecursorMz);
+                spectrum.AppendLine(" RT=" + RetentionTime);
+                spectrum.Append("Num alpha peaks: " + MatchedFragmentIons.Count);
+                spectrum.AppendLine(", Num beta peaks: " + BetaPeptideSpectrum.MatchedFragmentIons.Count);
+
+                double maxIntensity = Math.Max(MatchedFragmentIons.Max(b => b.Intensity),
+                    BetaPeptideSpectrum.MatchedFragmentIons.Max(s => s.Intensity));
+
+                foreach (MatchedFragmentIon matchedIon in MatchedFragmentIons)
+                {
+                    double intensityFraction = matchedIon.Intensity / maxIntensity;
+
+                    string neutralLoss = null;
+                    if (matchedIon.NeutralTheoreticalProduct.NeutralLoss != 0)
+                    {
+                        neutralLoss = "-" + matchedIon.NeutralTheoreticalProduct.NeutralLoss;
+                    }
+
+                    spectrum.AppendLine(matchedIon.Mz + "\t" + intensityFraction + "\t" + "\"" +
+                        matchedIon.NeutralTheoreticalProduct.ProductType.ToString() +
+                        matchedIon.NeutralTheoreticalProduct.FragmentNumber.ToString() + "^" +
+                        matchedIon.Charge + neutralLoss + "/" + 0 + "ppm" + "\"");
+                }
+
+                foreach (MatchedFragmentIon matchedIon in BetaPeptideSpectrum.MatchedFragmentIons)
+                {
+                    double intensityFraction = matchedIon.Intensity / maxIntensity;
+
+                    string neutralLoss = null;
+                    if (matchedIon.NeutralTheoreticalProduct.NeutralLoss != 0)
+                    {
+                        neutralLoss = "-" + matchedIon.NeutralTheoreticalProduct.NeutralLoss;
+                    }
+
+                    spectrum.AppendLine(matchedIon.Mz + "\t" + intensityFraction + "\t" + "\"" +
+                        matchedIon.NeutralTheoreticalProduct.ProductType.ToString() +
+                        matchedIon.NeutralTheoreticalProduct.FragmentNumber.ToString() + "^" +
+                        matchedIon.Charge + neutralLoss + "/" + 0 + "ppm" + "\"" +
+                        "\t" + "BetaPeptideIon");
+                }
+
+                return spectrum.ToString().Trim();
+            }
+        }
+    }

mzLib/Omics/SpectrumMatch/LibrarySpectrum.cs

@@ -15,7 +15,7 @@ public class LibrarySpectrum : MzSpectrum
         public List<MatchedFragmentIon> MatchedFragmentIons { get; set; }
         public bool IsDecoy { get; set; }
 
-        public string Name
+        public virtual string Name
         {
             get { return Sequence + "/" + ChargeState; }
         }

mzLib/Omics/SpectrumMatch/SpectrumMatchFromTsv.cs

@@ -156,7 +156,7 @@
         }
 
 
-        protected static List<MatchedFragmentIon> ReadFragmentIonsFromString(string matchedMzString, string matchedIntensityString, string peptideBaseSequence, string matchedMassErrorDaString = null)
+        protected static List<MatchedFragmentIon> ReadFragmentIonsFromString(string matchedMzString, string matchedIntensityString, string peptideBaseSequence, string matchedMassErrorDaString = null, bool isProtein = true)
         {
             List<MatchedFragmentIon> matchedIons = new List<MatchedFragmentIon>();
 
@@ -225,11 +225,14 @@
                         }
 
                         //get terminus
-                        if (TerminusSpecificProductTypes.ProductTypeToFragmentationTerminus.TryGetValue(productType,
-                                out terminus));
+
+                        if (isProtein)
+                            TerminusSpecificProductTypes.ProductTypeToFragmentationTerminus.TryGetValue(productType, out terminus);
+                        else
+                            terminus = Omics.Fragmentation.Oligo.TerminusSpecificProductTypes.GetRnaTerminusType(productType);
 
                         //get amino acid position
-                        aminoAcidPosition = terminus == FragmentationTerminus.C ?
+                        aminoAcidPosition = terminus is FragmentationTerminus.C or FragmentationTerminus.ThreePrime ?
                             peptideBaseSequence.Split('|')[0].Length - fragmentNumber :
                             fragmentNumber;
                     }
@@ -365,7 +368,7 @@
         {
             return FullSequence;
         }
-        public LibrarySpectrum ToLibrarySpectrum()
+        public virtual LibrarySpectrum ToLibrarySpectrum()
         {
             bool isDecoy = this.DecoyContamTarget == "D";
 

mzLib/Omics/SpectrumMatch/SpectrumMatchFromTsvHeader.cs

@@ -1,13 +1,5 @@
 namespace Omics.SpectrumMatch
 {
-    //for glcyo
-    public enum LocalizationLevel
-    {
-        Level1,
-        Level1b,
-        Level2,
-        Level3
-    }
     public class SpectrumMatchFromTsvHeader
     {
         // File and scan information
@@ -28,6 +20,7 @@ public class SpectrumMatchFromTsvHeader
         public const string BaseSequence = "Base Sequence";
         public const string FullSequence = "Full Sequence";
         public const string EssentialSequence = "Essential Sequence";
+        public const string UniqueSequence = "Unique Sequence"; //Used for crosslinked peptides
         public const string AmbiguityLevel = "Ambiguity Level";
         public const string SpectrumMatchCount = "Spectrum Match Count";
         public const string Mods = "Mods";

mzLib/Proteomics/PSM/PsmFromTsv.cs

@@ -3,6 +3,7 @@
 using System.Globalization;
 using System.IO;
 using System.Linq;
+using Easy.Common.Extensions;
 using Omics.Fragmentation;
 using Omics.SpectrumMatch;
 
@@ -34,6 +35,10 @@
         public int? BetaPeptideRank { get; }
         public List<MatchedFragmentIon> BetaPeptideMatchedIons { get; }
         public Dictionary<int, List<MatchedFragmentIon>> BetaPeptideChildScanMatchedIons { get; }
+        /// <summary>
+        /// If Crosslink, this contains the alpha and beta sequences. Otherwise, it contains the full sequence
+        /// </summary>
+        public string UniqueSequence { get; }
         public double? XLTotalScore { get; }
         public string ParentIons { get; }
 
@@ -135,6 +140,12 @@
                 ((spl[parsedHeader[SpectrumMatchFromTsvHeader.BetaPeptideMatchedIonsLabel]].StartsWith("{")) ? ReadChildScanMatchedIons(spl[parsedHeader[SpectrumMatchFromTsvHeader.BetaPeptideMatchedIonsLabel]].Trim(), spl[parsedHeader[SpectrumMatchFromTsvHeader.BetaPeptideMatchedIonIntensitiesLabel]].Trim(), BetaPeptideBaseSequence).First().Value : ReadFragmentIonsFromString(spl[parsedHeader[SpectrumMatchFromTsvHeader.BetaPeptideMatchedIonsLabel]].Trim(), spl[parsedHeader[SpectrumMatchFromTsvHeader.BetaPeptideMatchedIonIntensitiesLabel]].Trim(), BetaPeptideBaseSequence));
             XLTotalScore = (parsedHeader[SpectrumMatchFromTsvHeader.XLTotalScoreLabel] < 0) ? null : (double?)double.Parse(spl[parsedHeader[SpectrumMatchFromTsvHeader.XLTotalScoreLabel]].Trim(), CultureInfo.InvariantCulture);
             ParentIons = (parsedHeader[SpectrumMatchFromTsvHeader.ParentIonsLabel] < 0) ? null : spl[parsedHeader[SpectrumMatchFromTsvHeader.ParentIonsLabel]].Trim();
+            // This ensures backwards compatibility with old Crosslink Search Results
+            // This works because the alpha and beta peptide full sequences are written to tsv with their crosslink site included (e.g., PEPTIDEK(4))
+            if (UniqueSequence == null && BetaPeptideFullSequence != null)
+            {
+                UniqueSequence = FullSequence + BetaPeptideFullSequence;
+            }
 
             // child scan matched ions for xlink and glyco. we are getting them all above and then deleting primary scan ions here.
             ChildScanMatchedIons = (!spl[parsedHeader[SpectrumMatchFromTsvHeader.MatchedIonMzRatios]].StartsWith("{")) ? null : ReadChildScanMatchedIons(spl[parsedHeader[SpectrumMatchFromTsvHeader.MatchedIonMzRatios]].Trim(), spl[parsedHeader[SpectrumMatchFromTsvHeader.MatchedIonIntensities]].Trim(), BaseSeq);
@@ -152,18 +163,30 @@
             }
 
             //For Glyco            
-            GlycanMass = (parsedHeader[SpectrumMatchFromTsvHeader.GlycanMass] < 0) ? null : (double?)double.Parse(spl[parsedHeader[SpectrumMatchFromTsvHeader.GlycanMass]], CultureInfo.InvariantCulture);
-            GlycanComposition = (parsedHeader[SpectrumMatchFromTsvHeader.GlycanComposition] < 0) ? null : spl[parsedHeader[SpectrumMatchFromTsvHeader.GlycanComposition]];
-            GlycanStructure = (parsedHeader[SpectrumMatchFromTsvHeader.GlycanStructure] < 0) ? null : spl[parsedHeader[SpectrumMatchFromTsvHeader.GlycanStructure]];
-            var localizationLevel = (parsedHeader[SpectrumMatchFromTsvHeader.GlycanLocalizationLevel] < 0) ? null : spl[parsedHeader[SpectrumMatchFromTsvHeader.GlycanLocalizationLevel]];
-            if (localizationLevel != null)
+            try // Try is so that glyco and non-glyco psms can be read from the same file
             {
-                if (localizationLevel.Equals("NA"))
-                    GlycanLocalizationLevel = null;
-                else
-                    GlycanLocalizationLevel = (LocalizationLevel)Enum.Parse(typeof(LocalizationLevel), localizationLevel);
+                GlycanMass = (parsedHeader[PsmTsvHeader_Glyco.GlycanMass] < 0) ? null : (double?)double.Parse(spl[parsedHeader[PsmTsvHeader_Glyco.GlycanMass]], CultureInfo.InvariantCulture);
+                GlycanComposition = (parsedHeader[PsmTsvHeader_Glyco.GlycanComposition] < 0) ? null : spl[parsedHeader[PsmTsvHeader_Glyco.GlycanComposition]];
+                GlycanStructure = (parsedHeader[PsmTsvHeader_Glyco.GlycanStructure] < 0) ? null : spl[parsedHeader[PsmTsvHeader_Glyco.GlycanStructure]];
+                var localizationLevel = (parsedHeader[PsmTsvHeader_Glyco.GlycanLocalizationLevel] < 0) ? null : spl[parsedHeader[PsmTsvHeader_Glyco.GlycanLocalizationLevel]];
+                if (localizationLevel != null)
+                {
+                    if (localizationLevel.Equals("NA"))
+                        GlycanLocalizationLevel = null;
+                    else
+                        GlycanLocalizationLevel = (LocalizationLevel)Enum.Parse(typeof(LocalizationLevel), localizationLevel);
+                }
+                LocalizedGlycan = (parsedHeader[PsmTsvHeader_Glyco.LocalizedGlycan] < 0) ? null : spl[parsedHeader[PsmTsvHeader_Glyco.LocalizedGlycan]];
+
+            }
+            catch
+            {
+                GlycanMass = null;
+                GlycanComposition = null;
+                GlycanStructure = null;
+                GlycanLocalizationLevel = null;
+                LocalizedGlycan = null;
             }
-            LocalizedGlycan = (parsedHeader[SpectrumMatchFromTsvHeader.LocalizedGlycan] < 0) ? null : spl[parsedHeader[SpectrumMatchFromTsvHeader.LocalizedGlycan]];
         }
 
         /// <summary>
@@ -259,6 +282,38 @@
             LocalizedGlycan = psm.LocalizedGlycan;
         }
 
-
+        /// <summary>
+        /// Override library spectrum for cross link library spectrum implict conversion
+        /// </summary>
+        /// <returns></returns>
+        public override LibrarySpectrum ToLibrarySpectrum()
+        {
+            bool isDecoy = this.DecoyContamTarget == "D";
+
+            List<MatchedFragmentIon> fragments = new List<MatchedFragmentIon>();
+
+            double matchedIonIntensitySum = Math.Max(1.0, this.MatchedIons.Select(i => i.Intensity).Sum());
+
+            foreach (MatchedFragmentIon ion in this.MatchedIons)
+            {
+                Product product = new Product(ion.NeutralTheoreticalProduct.ProductType, ion.NeutralTheoreticalProduct.Terminus, ion.NeutralTheoreticalProduct.NeutralMass, ion.NeutralTheoreticalProduct.FragmentNumber, ion.NeutralTheoreticalProduct.AminoAcidPosition, ion.NeutralTheoreticalProduct.NeutralLoss);
+                fragments.Add(new MatchedFragmentIon(product, ion.Mz, ion.Intensity / matchedIonIntensitySum, ion.Charge));
+            }
+            double retentionTime = RetentionTime ?? -1;
+
+            if (BetaPeptideMatchedIons.IsNotNullOrEmpty())
+            {
+                List<MatchedFragmentIon> betaFragments = new();
+                foreach (var ion in BetaPeptideMatchedIons)
+                {
+                    Product product = new Product(ion.NeutralTheoreticalProduct.ProductType, ion.NeutralTheoreticalProduct.Terminus, ion.NeutralTheoreticalProduct.NeutralMass, ion.NeutralTheoreticalProduct.FragmentNumber, ion.NeutralTheoreticalProduct.AminoAcidPosition, ion.NeutralTheoreticalProduct.NeutralLoss);
+                    betaFragments.Add(new MatchedFragmentIon(product, ion.Mz, ion.Intensity / matchedIonIntensitySum, ion.Charge));
+                }
+                string uniqueSequence = UniqueSequence ?? FullSequence + BetaPeptideFullSequence;
+                return new CrosslinkLibrarySpectrum(uniqueSequence, PrecursorMz, PrecursorCharge, fragments, retentionTime, betaFragments);
+            }
+
+            return (new(this.FullSequence, this.PrecursorMz, this.PrecursorCharge, fragments, retentionTime, isDecoy));
+        }
     }
 }

mzLib/Proteomics/PSM/PsmTsvHeader_Glyco.cs

@@ -2,10 +2,10 @@
 {
     public enum LocalizationLevel
     {
-        Level1,
-        Level1b,
-        Level2,
-        Level3
+        Level1 = 0,
+        Level1b = 1,
+        Level2 = 2,
+        Level3 = 3
     }
     public class PsmTsvHeader_Glyco
     {

mzLib/Readers/Readers.csproj

@@ -17,6 +17,7 @@
 	<ItemGroup>
 		<ProjectReference Include="..\MassSpectrometry\MassSpectrometry.csproj" />
 		<ProjectReference Include="..\MzLibUtil\MzLibUtil.csproj" />
+		<ProjectReference Include="..\Transcriptomics\Transcriptomics.csproj" />
 		<ProjectReference Include="..\UsefulProteomicsDatabases\UsefulProteomicsDatabases.csproj" />
 	</ItemGroup>
 

mzLib/Readers/SearchResults/SpectrumMatchTsvReader.cs

@@ -53,14 +53,15 @@ public static List<SpectrumMatchFromTsv> ReadTsv(string filePath, out List<strin
                 {
                     switch (filePath.ParseFileType())
                     {
+                        //case SupportedFileType.osmtsv:
+                        //    psms.Add(new OsmFromTsv(line, Split, parsedHeader));
+                        //    break;
+
                         case SupportedFileType.psmtsv:
+                        case SupportedFileType.IntralinkResults:
+                        default:
                             psms.Add(new PsmFromTsv(line, Split, parsedHeader));
                             break;
-
-                        // TODO: Create an osmtsv case when transcriptomics is merged
-
-                        default:
-                            throw new ArgumentOutOfRangeException();
                     }
                 }
                 catch (Exception e)
@@ -123,8 +124,8 @@ public static Dictionary<string, int> ParseHeader(string header)
                 parsedHeader.Add(SpectrumMatchFromTsvHeader.Name, Array.IndexOf(spl, SpectrumMatchFromTsvHeader.Name));
                 parsedHeader.Add(SpectrumMatchFromTsvHeader.Description, Array.IndexOf(spl, SpectrumMatchFromTsvHeader.Description));
                 parsedHeader.Add(SpectrumMatchFromTsvHeader.StartAndEndResiduesInFullSequence, Array.IndexOf(spl, SpectrumMatchFromTsvHeader.StartAndEndResiduesInFullSequence));
-                parsedHeader.Add(SpectrumMatchFromTsvHeader.NextResidue, Array.IndexOf(spl, SpectrumMatchFromTsvHeader.PreviousResidue));
-                parsedHeader.Add(SpectrumMatchFromTsvHeader.PreviousResidue, Array.IndexOf(spl, SpectrumMatchFromTsvHeader.NumExperimentalPeaks));
+                parsedHeader.Add(SpectrumMatchFromTsvHeader.NextResidue, Array.IndexOf(spl, SpectrumMatchFromTsvHeader.NextResidue));
+                parsedHeader.Add(SpectrumMatchFromTsvHeader.PreviousResidue, Array.IndexOf(spl, SpectrumMatchFromTsvHeader.PreviousResidue));
             }
             else
             {
@@ -134,8 +135,8 @@ public static Dictionary<string, int> ParseHeader(string header)
                 parsedHeader.Add(SpectrumMatchFromTsvHeader.Name, Array.IndexOf(spl, SpectrumMatchFromTsvHeader.ProteinName));
                 parsedHeader.Add(SpectrumMatchFromTsvHeader.Description, Array.IndexOf(spl, SpectrumMatchFromTsvHeader.PeptideDescription));
                 parsedHeader.Add(SpectrumMatchFromTsvHeader.StartAndEndResiduesInFullSequence, Array.IndexOf(spl, SpectrumMatchFromTsvHeader.StartAndEndResiduesInProtein));
-                parsedHeader.Add(SpectrumMatchFromTsvHeader.NextResidue, Array.IndexOf(spl, SpectrumMatchFromTsvHeader.PreviousAminoAcid));
-                parsedHeader.Add(SpectrumMatchFromTsvHeader.PreviousResidue, Array.IndexOf(spl, SpectrumMatchFromTsvHeader.NextAminoAcid));
+                parsedHeader.Add(SpectrumMatchFromTsvHeader.NextResidue, Array.IndexOf(spl, SpectrumMatchFromTsvHeader.NextAminoAcid));
+                parsedHeader.Add(SpectrumMatchFromTsvHeader.PreviousResidue, Array.IndexOf(spl, SpectrumMatchFromTsvHeader.PreviousAminoAcid));
             }
 
             parsedHeader.Add(SpectrumMatchFromTsvHeader.GeneName, Array.IndexOf(spl, SpectrumMatchFromTsvHeader.GeneName));