IEquality hot fix hot fix (#819)

* Added parent accession check back to PeptideWithSetMods * Added Test * oopsies * Refactor Equals methods and add IEquatable regions Removed Equals from IBioPolymerWithSetMods interface in Omics. Updated Equals in PeptideWithSetModifications to check OneBasedStartResidue and Parent?.Accession. Updated Equals in OligoWithSetMods to check FullSequence and DigestionParams?.DigestionAgent. Added IEquatable regions to PeptideWithSetModifications and OligoWithSetMods. Organized Equals methods with #region IEquatable directives. * workflow edit * Even more equality tetsts * adjusted oligo with set mods * ugh --------- Co-authored-by: Nic Bollis <[email protected]> Co-authored-by: Alex <[email protected]>
smith-chem-wisc · Dec 18, 2024 · dc44773 · dc44773
1 parent 90bd259
commit dc44773
Show file tree

Hide file tree

Showing 7 changed files with 131 additions and 45 deletions.
diff --git a/mzLib/Omics/IBioPolymerWithSetMods.cs b/mzLib/Omics/IBioPolymerWithSetMods.cs
@@ -44,29 +44,6 @@ public interface IBioPolymerWithSetMods : IHasChemicalFormula, IEquatable<IBioPo
         char this[int zeroBasedIndex] => BaseSequence[zeroBasedIndex];
         IBioPolymer Parent { get; }
 
-        /// <summary>
-        /// Default Equals Method for IBioPolymerWithSetMods
-        /// </summary>
-        /// <param name="other"></param>
-        /// <returns></returns>
-        /// <remarks>
-        /// Different parent but same sequence and digestion condition => are equal
-        /// Different Digestion agent but same sequence => are not equal (this is for multi-protease analysis in MetaMorpheus)
-        /// </remarks>
-        bool IEquatable<IBioPolymerWithSetMods>.Equals(IBioPolymerWithSetMods? other)
-        {
-            if (other is null) return false;
-            if (ReferenceEquals(this, other)) return true;
-            if (other.GetType() != GetType()) return false;
-
-            // for those constructed from sequence and mods only
-            if (Parent is null && other.Parent is null)
-                return FullSequence.Equals(other.FullSequence);
-
-            return FullSequence == other.FullSequence
-                && Equals(DigestionParams?.DigestionAgent, other.DigestionParams?.DigestionAgent);
-        }
-
         public void Fragment(DissociationType dissociationType, FragmentationTerminus fragmentationTerminus,
             List<Product> products);
 

diff --git a/mzLib/Proteomics/ProteolyticDigestion/PeptideWithSetModifications.cs b/mzLib/Proteomics/ProteolyticDigestion/PeptideWithSetModifications.cs
@@ -16,7 +16,7 @@ namespace Proteomics.ProteolyticDigestion
     [Serializable]
     public class PeptideWithSetModifications : ProteolyticPeptide, IBioPolymerWithSetMods, IEquatable<PeptideWithSetModifications>
     {
-        public string FullSequence { get; private set; } //sequence with modifications
+        public string FullSequence { get; init; } //sequence with modifications
         public int NumFixedMods { get; }
         // Parameter to store the full sequence of the corresponding Target or Decoy peptide
         // If the peptide in question is a decoy, this pairs it to the target it was generated from
@@ -884,6 +884,11 @@ public override string ToString()
             return FullSequence + string.Join("\t", AllModsOneIsNterminus.Select(m => m.ToString()));
         }
 
+        #region IEquatable
+
+        /// <summary>
+        /// Peptides are equal if they have the same full sequence, parent, and digestion agent
+        /// </summary>
         public override bool Equals(object obj)
         {
             if (obj is PeptideWithSetModifications peptide)
@@ -893,12 +898,29 @@ public override bool Equals(object obj)
             return false;
         }
 
+        /// <summary>
+        /// Peptides are equal if they have the same full sequence, parent, and digestion agent
+        /// </summary>
+        public bool Equals(IBioPolymerWithSetMods other) => Equals(other as PeptideWithSetModifications);
+
+        /// <summary>
+        /// Peptides are equal if they have the same full sequence, parent, and digestion agent
+        /// </summary>
         public bool Equals(PeptideWithSetModifications other)
         {
-            // interface equals first because it does null and reference checks
-            return (this as IBioPolymerWithSetMods).Equals(other)
+            if (other is null) return false;
+            if (ReferenceEquals(this, other)) return true;
+            if (other.GetType() != GetType()) return false;
+
+            // for those constructed from sequence and mods only
+            if (Parent is null && other.Parent is null)
+                return FullSequence.Equals(other.FullSequence);
+
+            return FullSequence == other.FullSequence
+                   && Equals(DigestionParams?.DigestionAgent, other.DigestionParams?.DigestionAgent)
+                   // These last two are important for parsimony in MetaMorpheus
                    && OneBasedStartResidue == other!.OneBasedStartResidue
-                   && Equals(Parent, other.Parent);
+                   && Equals(Parent?.Accession, other.Parent?.Accession); 
         }
 
         public override int GetHashCode()
@@ -917,6 +939,8 @@ public override int GetHashCode()
             return hash.ToHashCode();
         }
 
+        #endregion
+
         /// <summary>
         /// This should be run after deserialization of a PeptideWithSetModifications, in order to set its Protein and Modification objects, which were not serialized
         /// </summary>

diff --git a/mzLib/Test/TestPeptideWithSetMods.cs b/mzLib/Test/TestPeptideWithSetMods.cs
@@ -71,6 +71,8 @@ public static void TestPeptideOligoEquality()
             Assert.That(!peptide.Equals(oligo));
             Assert.That(!((IBioPolymerWithSetMods)oligo).Equals(peptide));
             Assert.That(!((IBioPolymerWithSetMods)peptide).Equals(oligo));
+            Assert.That(!((object)oligo).Equals(peptide));
+            Assert.That(!((object)peptide).Equals(oligo));
         }
 
         [Test]
@@ -1194,6 +1196,45 @@ public static void TestPeptideWithSetModsNoParentProtein()
             Assert.AreEqual('-', last.NextResidue);
         }
 
+        [Test]
+        public static void TestPeptideWithSetModsEquals()
+        {
+            // Create two proteins
+            Protein protein1 = new Protein("SEQUENCEK", "accession1");
+            Protein protein2 = new Protein("SEQUENCEK", "accession2");
+
+            // Create digestion parameters
+            DigestionParams digestionParams = new DigestionParams(protease: "trypsin", maxMissedCleavages: 0, initiatorMethionineBehavior: InitiatorMethionineBehavior.Retain);
+
+            // Digest the proteins to get peptides
+            PeptideWithSetModifications peptide1 = protein1.Digest(digestionParams, new List<Modification>(), new List<Modification>()).First();
+            PeptideWithSetModifications peptide2 = protein2.Digest(digestionParams, new List<Modification>(), new List<Modification>()).First();
+
+            // Test equality - same peptide
+            Assert.IsTrue(peptide1.Equals(peptide1));
+
+            // different peptide
+            Assert.IsTrue(!peptide1.Equals(peptide2));
+            Assert.IsTrue(!peptide1.Equals((object)peptide2));
+            Assert.IsTrue(!peptide1.Equals((IBioPolymerWithSetMods)peptide2));
+            Assert.AreNotEqual(peptide1.GetHashCode(), peptide2.GetHashCode());
+
+            // Test inequality with different start residue
+            PeptideWithSetModifications peptide3 = new PeptideWithSetModifications(protein1, digestionParams, 2, 9, CleavageSpecificity.Full, "", 0, new Dictionary<int, Modification>(), 0);
+            Assert.IsFalse(peptide1.Equals(peptide3));
+
+            // Test inequality with different parent accession
+            PeptideWithSetModifications peptide4 = new PeptideWithSetModifications(protein2, digestionParams, 1, 9, CleavageSpecificity.Full, "", 0, new Dictionary<int, Modification>(), 0);
+            Assert.IsFalse(peptide1.Equals(peptide4));
+
+            // all fail on null
+            Assert.That(!peptide1.Equals(null));
+            Assert.That(!peptide1.Equals((object)null));
+            Assert.That(!peptide1.Equals((PeptideWithSetModifications)null));
+        }
+
+
+
         [Test]
         public static void TestIBioPolymerWithSetModsModificationFromFullSequence()
         {

diff --git a/mzLib/Test/Transcriptomics/TestOligoWithSetMods.cs b/mzLib/Test/Transcriptomics/TestOligoWithSetMods.cs
@@ -91,43 +91,55 @@ public static void TestEquality()
                 .Digest(new RnaDigestionParams(), [], [])
                 .ElementAt(1);
 
-            Assert.That(oligoWithSetMods, Is.EqualTo(oligoWithSetMods2));
+            // same oligos
+            Assert.That(oligoWithSetMods.Equals(oligoWithSetMods2));
+            Assert.That(oligoWithSetMods.Equals((object)oligoWithSetMods2));
+            Assert.That(oligoWithSetMods.Equals((OligoWithSetMods)oligoWithSetMods2));
+            Assert.That(oligoWithSetMods.Equals(oligoWithSetMods));
+            Assert.That(oligoWithSetMods.Equals((object)oligoWithSetMods));
+            Assert.That(oligoWithSetMods.Equals((OligoWithSetMods)oligoWithSetMods));
             Assert.That(oligoWithSetMods.GetHashCode(), Is.EqualTo(oligoWithSetMods2.GetHashCode()));
-            Assert.That(oligoWithSetMods.Equals((object)oligoWithSetMods2)); // Test the Equals(Object obj) method
+
+            // all fail on null
             Assert.That(!oligoWithSetMods2.Equals(null));
+            Assert.That(!oligoWithSetMods2.Equals((object)null));
+            Assert.That(!oligoWithSetMods2.Equals((OligoWithSetMods)null));
 
             // Null parent checks
             oligoWithSetMods = new(oligoWithSetMods.FullSequence, modDict.ToDictionary(p => p.Value.First().IdWithMotif, p => p.Value.First()));
             oligoWithSetMods2 = new OligoWithSetMods(oligoWithSetMods.FullSequence, modDict.ToDictionary(p => p.Value.First().IdWithMotif, p => p.Value.First()));
             var oligoWithSetMods3 = new OligoWithSetMods(oligoWithSetMods.FullSequence + "AGAUA", modDict.ToDictionary(p => p.Value.First().IdWithMotif, p => p.Value.First()));
 
-            Assert.That(oligoWithSetMods, Is.EqualTo(oligoWithSetMods2));
-            Assert.That(oligoWithSetMods, Is.EqualTo((object)oligoWithSetMods2));
-            Assert.That(oligoWithSetMods, Is.EqualTo((OligoWithSetMods)oligoWithSetMods2));
-            Assert.That(oligoWithSetMods, Is.Not.EqualTo(oligoWithSetMods3));
-            Assert.That(oligoWithSetMods, Is.Not.EqualTo((object)oligoWithSetMods3));
-            Assert.That(oligoWithSetMods, Is.Not.EqualTo((IBioPolymerWithSetMods)oligoWithSetMods3));
+            // same oligo null parent
+            Assert.That(oligoWithSetMods.Equals(oligoWithSetMods2));
+            Assert.That(oligoWithSetMods.Equals((object)oligoWithSetMods2));
+            Assert.That(oligoWithSetMods.Equals((OligoWithSetMods)oligoWithSetMods2));
+
+            // different oligo null parent
+            Assert.That(!oligoWithSetMods.Equals(oligoWithSetMods3));
+            Assert.That(!oligoWithSetMods.Equals((object)oligoWithSetMods3));
+            Assert.That(!oligoWithSetMods.Equals((IBioPolymerWithSetMods)oligoWithSetMods3));
         }
 
         [Test]
         [TestCase("GUACUG", "GUACUGGUACUG", "RNase A")]
         [TestCase("GUAGGAG", "GUAGCAG", "RNase A")]
-        public static void TestEquality_DifferentParentSameDigestionProduct(string sequence1, string sequence2, string enzyme)
+        public static void TestInequality_DifferentParentSameDigestionProduct(string sequence1, string sequence2, string enzyme)
         {
             var digestionParams = new RnaDigestionParams(rnase: enzyme, minLength: 1, maxMissedCleavages: 0);
 
-            var oligo1 = new RNA(sequence1)
+            var oligo1 = new RNA(sequence1, "", "rna1", "", "")
                 .Digest(digestionParams, [], [])
                 .First();
 
-            var oligo2 = new RNA(sequence2)
+            var oligo2 = new RNA(sequence2, "", "rna3", "", "")
                 .Digest(digestionParams, [], [])
                 .First();
 
-            Assert.That(oligo1, Is.EqualTo(oligo2));
+            Assert.That(oligo1, Is.Not.EqualTo(oligo2));
             Assert.That(oligo1.Equals(oligo1));
-            Assert.That(oligo1, Is.EqualTo((object)oligo2));
-            Assert.That(oligo1.GetHashCode(), Is.EqualTo(oligo2.GetHashCode()));
+            Assert.That(oligo1, Is.Not.EqualTo((object)oligo2));
+            Assert.That(oligo1.GetHashCode(), Is.Not.EqualTo(oligo2.GetHashCode()));
         }
 
         /// <summary>

diff --git a/mzLib/Transcriptomics/Digestion/OligoWithSetMods.cs b/mzLib/Transcriptomics/Digestion/OligoWithSetMods.cs
@@ -215,6 +215,11 @@ public void Fragment(DissociationType dissociationType, FragmentationTerminus fr
                     products.AddRange(GetNeutralFragments(type, sequence));
         }
 
+        #region IEquatable
+
+        /// <summary>
+        /// Oligos are equal if they have the same full sequence, parent, and digestion agent, and terminal caps
+        /// </summary>
         public override bool Equals(object? obj)
         {
             if (obj is OligoWithSetMods oligo)
@@ -224,9 +229,31 @@ public override bool Equals(object? obj)
             return false;
         }
 
+        /// <summary>
+        /// Oligos are equal if they have the same full sequence, parent, and digestion agent, and terminal caps
+        /// </summary>
+        public bool Equals(IBioPolymerWithSetMods? other) => Equals(other as OligoWithSetMods);
+
+        /// <summary>
+        /// Oligos are equal if they have the same full sequence, parent, and digestion agent, and terminal caps
+        /// </summary>
         public bool Equals(OligoWithSetMods? other)
         {
-            return (this as IBioPolymerWithSetMods).Equals(other);
+            if (other is null) return false;
+            if (ReferenceEquals(this, other)) return true;
+            if (other.GetType() != GetType()) return false;
+
+            // for those constructed from sequence and mods only
+            if (Parent is null && other.Parent is null)
+                return FullSequence.Equals(other.FullSequence);
+
+            return FullSequence == other.FullSequence
+                   && Equals(DigestionParams?.DigestionAgent, other.DigestionParams?.DigestionAgent)
+                   && _fivePrimeTerminus.Equals(other._fivePrimeTerminus)
+                   && _threePrimeTerminus.Equals(other._threePrimeTerminus)
+                   // These last two are important for parsimony in MetaMorpheus
+                   && OneBasedStartResidue == other!.OneBasedStartResidue
+                   && Equals(Parent?.Accession, other.Parent?.Accession);
         }
 
         public override int GetHashCode()
@@ -242,9 +269,13 @@ public override int GetHashCode()
             {
                 hash.Add(DigestionParams.DigestionAgent);
             }
+            hash.Add(FivePrimeTerminus);
+            hash.Add(ThreePrimeTerminus);
             return hash.ToHashCode();
         }
 
+        #endregion
+
         /// <summary>
         /// Generates theoretical internal fragments for given dissociation type for this peptide. 
         /// The "products" parameter is filled with these fragments.

diff --git a/mzLib/Transcriptomics/RNA.cs b/mzLib/Transcriptomics/RNA.cs
@@ -23,7 +23,7 @@ public RNA(string sequence, IHasChemicalFormula? fivePrimeTerm = null, IHasChemi
         /// </summary>
         /// <param name="sequence"></param>
         /// <param name="name"></param>
-        /// <param name="identifier"></param>
+        /// <param name="accession"></param>
         /// <param name="organism"></param>
         /// <param name="databaseFilePath"></param>
         /// <param name="fivePrimeTerminus"></param>
@@ -33,12 +33,12 @@ public RNA(string sequence, IHasChemicalFormula? fivePrimeTerm = null, IHasChemi
         /// <param name="isDecoy"></param>
         /// <param name="geneNames"></param>
         /// <param name="databaseAdditionalFields"></param>
-        public RNA(string sequence, string name, string identifier, string organism, string databaseFilePath,
+        public RNA(string sequence, string name, string accession, string organism, string databaseFilePath,
             IHasChemicalFormula? fivePrimeTerminus = null, IHasChemicalFormula? threePrimeTerminus = null,
             IDictionary<int, List<Modification>>? oneBasedPossibleModifications = null,
             bool isContaminant = false, bool isDecoy = false, List<Tuple<string, string>> geneNames = null,
             Dictionary<string, string>? databaseAdditionalFields = null)
-            : base(sequence, name, identifier, organism, databaseFilePath, fivePrimeTerminus, threePrimeTerminus,
+            : base(sequence, name, accession, organism, databaseFilePath, fivePrimeTerminus, threePrimeTerminus,
                 oneBasedPossibleModifications, isContaminant, isDecoy, geneNames, databaseAdditionalFields)
         {
 

diff --git a/mzLib/mzLib.sln.DotSettings b/mzLib/mzLib.sln.DotSettings
@@ -8,6 +8,7 @@
 	<s:Boolean x:Key="/Default/UserDictionary/Words/=Modomics/@EntryIndexedValue">True</s:Boolean>
 	<s:Boolean x:Key="/Default/UserDictionary/Words/=Nucleolytic/@EntryIndexedValue">True</s:Boolean>
 	<s:Boolean x:Key="/Default/UserDictionary/Words/=Oligo/@EntryIndexedValue">True</s:Boolean>
+	<s:Boolean x:Key="/Default/UserDictionary/Words/=Oligos/@EntryIndexedValue">True</s:Boolean>
 	<s:Boolean x:Key="/Default/UserDictionary/Words/=Prsm/@EntryIndexedValue">True</s:Boolean>
 	<s:Boolean x:Key="/Default/UserDictionary/Words/=Regexes/@EntryIndexedValue">True</s:Boolean>
 	<s:Boolean x:Key="/Default/UserDictionary/Words/=Toppic/@EntryIndexedValue">True</s:Boolean>