Digestion: Fixed mod terminal fix and variable mod ordering

mzLib/Omics/Digestion/DigestionProduct.cs

@@ -1,11 +1,11 @@
-using MzLibUtil;
+using MzLibUtil;
 using Omics.Modifications;
 
 namespace Omics.Digestion
 {
     public abstract class DigestionProduct
     {
-        protected static readonly DictionaryPool<int, List<Modification>> DictionaryPool = new();
+        protected static readonly DictionaryPool<int, SortedSet<Modification>> DictionaryPool = new();
         protected static readonly DictionaryPool<int, Modification> FixedModDictionaryPool = new(8);
 
         protected string _baseSequence;
@@ -54,29 +54,30 @@ protected DigestionProduct(IBioPolymer parent, int oneBasedStartResidue, int one
         /// Then, it iterates through all possible numbers of modifications and generates the corresponding modification patterns.
         /// The returned dictionary is then appended with fixed modifications and used to construct a peptide with set mods
         /// </remarks>
-        protected static IEnumerable<Dictionary<int, Modification>> GetVariableModificationPatterns(Dictionary<int, List<Modification>> possibleVariableModifications, int maxModsForPeptide, int peptideLength)
+        protected static IEnumerable<Dictionary<int, Modification>> GetVariableModificationPatterns(Dictionary<int, SortedSet<Modification>> possibleVariableModifications, int maxModsForPeptide, int peptideLength)
         {
             if (possibleVariableModifications.Count <= 0) 
                 yield break;
 
             int[] baseVariableModificationPattern = new int[peptideLength + 4];
             int totalAvailableMods = possibleVariableModifications.Values.Sum(modList => modList?.Count ?? 0);
             int maxVariableMods = Math.Min(totalAvailableMods, maxModsForPeptide);
+            var variableModKvpList = possibleVariableModifications.ToList();
 
             for (int variable_modifications = 0; variable_modifications <= maxVariableMods; variable_modifications++)
             {
-                foreach (int[] variable_modification_pattern in GetVariableModificationPatternsRecursive(possibleVariableModifications.ToList(),
+                foreach (int[] variable_modification_pattern in GetVariableModificationPatternsRecursive(variableModKvpList,
                              possibleVariableModifications.Count - variable_modifications, baseVariableModificationPattern, 0))
                 {
                     // use modification pattern to construct a dictionary of modifications for the peptide
                     var modificationPattern = new Dictionary<int, Modification>(possibleVariableModifications.Count);
 
-                    foreach (KeyValuePair<int, List<Modification>> kvp in possibleVariableModifications)
+                    foreach (var variableModSet in possibleVariableModifications)
                     {
-                        int modIndex = variable_modification_pattern[kvp.Key] - 1;
+                        int modIndex = variable_modification_pattern[variableModSet.Key] - 1;
                         if (modIndex >= 0)
                         {
-                            modificationPattern.Add(kvp.Key, kvp.Value[modIndex]);
+                            modificationPattern.Add(variableModSet.Key, variableModSet.Value.ElementAt(modIndex));
                         }
                     }
 
@@ -162,13 +163,13 @@ protected void PopulateFixedModsOneIsNorFivePrimeTerminus(int length,
         /// This method iterates through all variable modifications and assigns them to the appropriate positions in the peptide.
         /// It considers different location restrictions such as N-terminal, C-terminal, and anywhere within the peptide.
         /// </remarks>
-        protected void PopulateVariableModifications(List<Modification> allVariableMods, in Dictionary<int, List<Modification>> twoBasedDictToPopulate)
+        protected void PopulateVariableModifications(List<Modification> allVariableMods, in Dictionary<int, SortedSet<Modification>> twoBasedDictToPopulate)
         {
             int peptideLength = OneBasedEndResidue - OneBasedStartResidue + 1;
-            var pepNTermVariableMods = new List<Modification>();
+            var pepNTermVariableMods = new SortedSet<Modification>();
             twoBasedDictToPopulate.Add(1, pepNTermVariableMods);
 
-            var pepCTermVariableMods = new List<Modification>();
+            var pepCTermVariableMods = new SortedSet<Modification>();
             twoBasedDictToPopulate.Add(peptideLength + 2, pepCTermVariableMods);
 
             // VARIABLE MODS
@@ -187,7 +188,7 @@ protected void PopulateVariableModifications(List<Modification> allVariableMods,
                     {
                         if (!twoBasedDictToPopulate.TryGetValue(r + 2, out var residueVariableMods))
                         {
-                            residueVariableMods = new List<Modification>() { variableModification };
+                            residueVariableMods = new SortedSet<Modification>() { variableModification };
                             twoBasedDictToPopulate.Add(r + 2, residueVariableMods);
                         }
                         else
@@ -232,7 +233,7 @@ protected void PopulateVariableModifications(List<Modification> allVariableMods,
                     {
                         if (!twoBasedDictToPopulate.TryGetValue(r + 2, out var residueVariableMods))
                         {
-                            residueVariableMods = new List<Modification>() { variableModification };
+                            residueVariableMods = new SortedSet<Modification>() { variableModification };
                             twoBasedDictToPopulate.Add(r + 2, residueVariableMods);
                         }
                         else
@@ -267,7 +268,6 @@ protected void AppendFixedModificationsToVariable(in Dictionary<int, Modificatio
             {
                 if (variableModPattern.ContainsKey(fixedModPattern.Key))
                     continue;
-
                 numFixedMods++;
                 variableModPattern.Add(fixedModPattern.Key, fixedModPattern.Value);
             }
@@ -287,7 +287,7 @@ protected void AppendFixedModificationsToVariable(in Dictionary<int, Modificatio
         /// This method uses recursion to generate all possible combinations of variable modifications for a given peptide.
         /// It considers both modified and unmodified residues and generates patterns accordingly.
         /// </remarks>
-        private static IEnumerable<int[]> GetVariableModificationPatternsRecursive(List<KeyValuePair<int, List<Modification>>> possibleVariableModifications,
+        private static IEnumerable<int[]> GetVariableModificationPatternsRecursive(List<KeyValuePair<int, SortedSet<Modification>>> possibleVariableModifications,
             int unmodifiedResiduesDesired, int[] variableModificationPattern, int index)
         {
             if (index < possibleVariableModifications.Count - 1)

mzLib/Omics/Modifications/Modification.cs

@@ -1,10 +1,5 @@
 using Chemistry;
 using MassSpectrometry;
-using Omics.Modifications;
-using System;
-using System.Collections.Generic;
-using System.Globalization;
-using System.Linq;
 using System.Text;
 
 namespace Omics.Modifications
@@ -13,7 +8,7 @@
     /// Represents a modification
     /// Mods.txt format was taken from https://ftp.uniprot.org/pub/databases/uniprot/current_release/knowledgebase/complete/docs/ptmlist.txt
     /// </summary>
-    public class Modification
+    public class Modification : IComparable<Modification>
     {
         public string IdWithMotif { get; private set; }
         public string OriginalId { get; private set; }
@@ -59,7 +54,7 @@
             }
         }
 
         public Modification(string _originalId = null, string _accession = null, string _modificationType = null, string _featureType = null,
             ModificationMotif _target = null, string _locationRestriction = "Unassigned.", ChemicalFormula _chemicalFormula = null,
             double? _monoisotopicMass = null, Dictionary<string, IList<string>> _databaseReference = null,
             Dictionary<string, IList<string>> _taxonomicRange = null, List<string> _keywords = null,
@@ -129,7 +124,7 @@
             }
         }
 
         public override bool Equals(object o)
         {
             Modification m = o as Modification;
             return o != null
@@ -299,5 +294,24 @@
 
             return sb.ToString();
         }
+
+
+        // Used in the sorted sets for variable mod generation to ensure that modifications are consistently ordered
+        // UniProt annotations also contain an evidence level. Future work could include this in the ordering of modifications for digestion. 
+        public int CompareTo(Modification? other)
+        {
+            if (other == null) return 1;
+
+            int idComparison = string.Compare(this.IdWithMotif, other.IdWithMotif, StringComparison.Ordinal);
+            if (idComparison != 0) return idComparison;
+
+            int typeComparison = string.Compare(this.ModificationType, other.ModificationType, StringComparison.Ordinal);
+            if (typeComparison != 0) return typeComparison;
+
+            int locRestrictionComparison = string.Compare(this.LocationRestriction, other.LocationRestriction, StringComparison.Ordinal);
+            if (locRestrictionComparison != 0) return locRestrictionComparison;
+
+            return Nullable.Compare(this.MonoisotopicMass, other.MonoisotopicMass);
+        }
     }
 }

mzLib/Omics/Modifications/ModificationLocalization.cs

@@ -32,9 +32,9 @@ public static bool ModFits(Modification attemptToLocalize, string sequence, int
             switch (attemptToLocalize.LocationRestriction)
             {
                 case "N-terminal." when bioPolymerOneBasedIndex > 2:
-                case "Peptide N-terminal." when digestionProductOneBasedIndex > 1:
+                case "Peptide N-terminal." when digestionProductOneBasedIndex > 1 || bioPolymerOneBasedIndex == 1:
                 case "C-terminal." when bioPolymerOneBasedIndex < sequence.Length:
-                case "Peptide C-terminal." when digestionProductOneBasedIndex < digestionProductLength:
+                case "Peptide C-terminal." when digestionProductOneBasedIndex < digestionProductLength || bioPolymerOneBasedIndex == sequence.Length:
                 case "5'-terminal." when bioPolymerOneBasedIndex > 2:
                 // first residue in oligo but not first in nucleic acid
                 case "Oligo 5'-terminal." when digestionProductOneBasedIndex > 1

mzLib/Omics/Modifications/ModificationMotif.cs

@@ -28,19 +28,6 @@ public static bool TryGetMotif(string motifString, out ModificationMotif motif)
             }
             return false;
         }
-        // Commented out by AVC on 4/5/23. Methods were unused and untested 
-        // since 2017. 
-        // public override bool Equals(object o)
-        // {
-        //     ModificationMotif m = o as ModificationMotif;
-        //     return m != null
-        //         && m.motifString == motifString;
-        // }
-        //
-        // public override int GetHashCode()
-        // {
-        //     return motifString.GetHashCode();
-        // }
 
         public override string ToString()
         {