Top down biomarker (#621)

* correct Within calculation

* update unit tests

* first correct top-down biomarker test

* remove unused test code

* quotable protease

* unquotable

* add full length proteoform proteolysis products for biomarker search with  unit tests

* new strategy for adding proteolysis products

* more complete biomarker addition

* works for protein.xml databases

* add unit test biomarkers with xml database

* clean up

* more clean up

* fix unit test

* adjust biomarker generation for three initiator methionine behaviors (cleave, retain, variable)

* make sure decoy biomarkers for standard xml proteolysis products are generated

* deal appropriately w/ n-terminal methionine cleavage

* last litle unit test tweaks

* delete unused code

* add biomarkers during protein database load

* greater clarity for biomarker proteolysis product entries

* change digest for top-down biomarker search to resolve issues in MM

* bro. shouldn't have been that hard

* cleanup

* add unit test

* don't need to cover unit tests with unit tests

* increase unit test coverage of protein.cs

* codemaid

* biomarkers on protein with no methionine

* unit test booster

* ACs comments

* more summary statemsn

* deleted mzlib.sln scarry

Co-authored-by: MICHAEL SHORTREED <mrshortreed@wisc.edu>

Author: trishorts

mzLib/Proteomics/Protein/Protein.cs

@@ -64,6 +64,7 @@ public Protein(string sequence, string accession, string organism = null, List<T
             DatabaseReferences = databaseReferences ?? new List<DatabaseReference>();
             DisulfideBonds = disulfideBonds ?? new List<DisulfideBond>();
             SpliceSites = spliceSites ?? new List<SpliceSite>();
+
             if (addBiomarkers)
             {
                 this.AddBiomarkers();
@@ -236,7 +237,7 @@ public string GetEnsemblFastaHeader()
         /// Gets peptides for digestion of a protein
         /// </summary>
         public IEnumerable<PeptideWithSetModifications> Digest(DigestionParams digestionParams, List<Modification> allKnownFixedModifications,
-            List<Modification> variableModifications, List<SilacLabel> silacLabels = null, (SilacLabel startLabel, SilacLabel endLabel)? turnoverLabels = null)
+            List<Modification> variableModifications, List<SilacLabel> silacLabels = null, (SilacLabel startLabel, SilacLabel endLabel)? turnoverLabels = null, bool topDownBiomarkerSearch = false)
         {
             //can't be null
             allKnownFixedModifications = allKnownFixedModifications ?? new List<Modification>();
@@ -252,7 +253,7 @@ public IEnumerable<PeptideWithSetModifications> Digest(DigestionParams digestion
             IEnumerable<ProteolyticPeptide> unmodifiedPeptides =
                 searchModeType == CleavageSpecificity.Semi ?
                 digestion.SpeedySemiSpecificDigestion(this) :
-                digestion.Digestion(this);
+                    digestion.Digestion(this, topDownBiomarkerSearch);
 
             if (digestionParams.KeepNGlycopeptide || digestionParams.KeepOGlycopeptide)
             {
@@ -717,18 +718,18 @@ public void AddBiomarkers(bool addFullProtein = true, bool addForEachOrigninalPr
                 AddIntactProteoformToProteolysisProducts(initiatorMethionineBehavior, minProductBaseSequenceLength);
                 if (addNterminalDigestionBiomarkers)
                 {
-                    AddBiomarkersToProteolysisProducts(1, BaseSequence.Length, true, false, initiatorMethionineBehavior, minProductBaseSequenceLength, lengthOfProteolysis, "full proteoform biomarker");
+                    AddBiomarkersToProteolysisProducts(1, BaseSequence.Length, true, false, initiatorMethionineBehavior, minProductBaseSequenceLength, lengthOfProteolysis, "full-length proteoform N-terminal digestion biomarker");
                 }
                 if (addCterminalDigestionBiomarkers)
                 {
-                    AddBiomarkersToProteolysisProducts(1, BaseSequence.Length, false, true, initiatorMethionineBehavior, minProductBaseSequenceLength, lengthOfProteolysis, "full proteoform biomarker");
+                    AddBiomarkersToProteolysisProducts(1, BaseSequence.Length, false, true, initiatorMethionineBehavior, minProductBaseSequenceLength, lengthOfProteolysis, "full-length proteoform C-terminal digestion biomarker");
                 }
             }
 
             if (addForEachOrigninalProteolysisProduct) // this does not include the original intact proteoform
             {
                 RemoveMethionineWhenAppropriateFromExistingProduts(initiatorMethionineBehavior);
-                List<ProteolysisProduct> existingProducts = ProteolysisProducts.Where(p => !p.Type.Contains("biomarker") && !p.Type.Contains("intact")).ToList();
+                List<ProteolysisProduct> existingProducts = ProteolysisProducts.Where(p => !p.Type.Contains("biomarker") && !p.Type.Contains("full-length proteoform")).ToList();
                 foreach (ProteolysisProduct product in existingProducts)
                 {
                     if (product.OneBasedBeginPosition.HasValue && product.OneBasedEndPosition.HasValue)
@@ -795,7 +796,7 @@ public void AddIntactProteoformToProteolysisProducts(InitiatorMethionineBehavior
                 //when it's variable, we don't have to add anything here, we'll get an additonal proteolysis product later.
                 if (BaseSequence.Length >= minProductBaseSequenceLength)
                 {
-                    _proteolysisProducts.Add(new ProteolysisProduct(1, BaseSequence.Length, "intact proteoform biomarker"));
+                    _proteolysisProducts.Add(new ProteolysisProduct(1, BaseSequence.Length, "full-length proteoform"));
                 }
             }
             else if (initiatorMethionineBehavior == InitiatorMethionineBehavior.Cleave)
@@ -804,14 +805,14 @@ public void AddIntactProteoformToProteolysisProducts(InitiatorMethionineBehavior
                 {
                     if (BaseSequence.Length - 1 >= minProductBaseSequenceLength)
                     {
-                        _proteolysisProducts.Add(new ProteolysisProduct(2, BaseSequence.Length, "intact proteoform biomarker"));
+                        _proteolysisProducts.Add(new ProteolysisProduct(2, BaseSequence.Length, "full-length proteoform"));
                     }
                 }
                 else
                 {
                     if (BaseSequence.Length >= minProductBaseSequenceLength)
                     {
-                        _proteolysisProducts.Add(new ProteolysisProduct(1, BaseSequence.Length, "intact proteoform biomarker"));
+                        _proteolysisProducts.Add(new ProteolysisProduct(1, BaseSequence.Length, "full-length proteoform"));
                     }
                 }
             }
@@ -822,16 +823,17 @@ public void AddIntactProteoformToProteolysisProducts(InitiatorMethionineBehavior
         /// This method allows for one missed cleavage between proteolysis products.
         /// </summary>
         /// <param name="minimumProductLength"></param>
+
         public void CleaveOnceBetweenProteolysisProducts(int minimumProductLength = 7)
         {
             List<int> cleavagePostions = new();
-            List<int> proteolysisProductEndPositions = _proteolysisProducts.Where(p => p.OneBasedEndPosition.HasValue).Select(p => p.OneBasedEndPosition.Value).ToList();
-
+            List<ProteolysisProduct> localProducts = _proteolysisProducts.Where(p => !p.Type.Contains("biomarker") && !p.Type.Contains("full-length proteoform")).ToList();
+            List<int> proteolysisProductEndPositions = localProducts.Where(p => p.OneBasedEndPosition.HasValue).Select(p => p.OneBasedEndPosition.Value).ToList();
             if (proteolysisProductEndPositions.Count > 0)
             {
                 foreach (int proteolysisProductEndPosition in proteolysisProductEndPositions)
                 {
-                    if (_proteolysisProducts.Any(p => p.OneBasedBeginPosition == (proteolysisProductEndPosition + 1)))
+                    if (localProducts.Any(p => p.OneBasedBeginPosition == (proteolysisProductEndPosition + 1)))
                     {
                         cleavagePostions.Add(proteolysisProductEndPosition);
                     }
@@ -844,6 +846,7 @@ public void CleaveOnceBetweenProteolysisProducts(int minimumProductLength = 7)
                 {
                     string leftType = $"N-terminal Portion of Singly Cleaved Protein(1-{position})";
                     ProteolysisProduct leftProduct = new(1, position, leftType);
+
                     //here we're making sure a product with these begin/end positions isn't already present
                     if (!_proteolysisProducts.Any(p => p.OneBasedBeginPosition == leftProduct.OneBasedBeginPosition && p.OneBasedEndPosition == leftProduct.OneBasedEndPosition))
                     {
@@ -855,6 +858,7 @@ public void CleaveOnceBetweenProteolysisProducts(int minimumProductLength = 7)
                 {
                     string rightType = $"C-terminal Portion of Singly Cleaved Protein({position + 1}-{BaseSequence.Length})";
                     ProteolysisProduct rightProduct = new(position + 1, BaseSequence.Length, rightType);
+
                     //here we're making sure a product with these begin/end positions isn't already present
                     if (!_proteolysisProducts.Any(p => p.OneBasedBeginPosition == rightProduct.OneBasedBeginPosition && p.OneBasedEndPosition == rightProduct.OneBasedEndPosition))
                     {

mzLib/Proteomics/ProteolyticDigestion/Protease.cs

@@ -13,15 +13,15 @@ public Protease(string name, CleavageSpecificity cleavageSpecificity, string psi
             PsiMsAccessionNumber = psiMSAccessionNumber;
             PsiMsName = psiMSName;
             DigestionMotifs = motifList ?? new List<DigestionMotif>();
-            CleavageMod = modDetails; 
+            CleavageMod = modDetails;
         }
 
         public string Name { get; }
         public CleavageSpecificity CleavageSpecificity { get; }
         public string PsiMsAccessionNumber { get; }
         public string PsiMsName { get; }
         public List<DigestionMotif> DigestionMotifs { get; }
-        public Modification CleavageMod {get; set;}
+        public Modification CleavageMod { get; set; }
 
         public override string ToString()
         {
@@ -38,7 +38,6 @@ public override int GetHashCode()
         {
             return (Name ?? "").GetHashCode();
         }
-               
 
         /// <summary>
         /// This method is used to determine cleavage specificity if the cleavage specificity is unknown
@@ -93,7 +92,7 @@ public CleavageSpecificity GetCleavageSpecificity(Protein protein, int startInde
         /// <param name="maxPeptideLength"></param>
         /// <returns></returns>
         internal List<ProteolyticPeptide> GetUnmodifiedPeptides(Protein protein, int maximumMissedCleavages, InitiatorMethionineBehavior initiatorMethionineBehavior,
-            int minPeptideLength, int maxPeptideLength, Protease specificProtease)
+            int minPeptideLength, int maxPeptideLength, Protease specificProtease, bool topDownBiomarkerSearch = false)
         {
             List<ProteolyticPeptide> peptides = new List<ProteolyticPeptide>();
 
@@ -112,7 +111,7 @@ internal List<ProteolyticPeptide> GetUnmodifiedPeptides(Protein protein, int max
             //top-down
             else if (CleavageSpecificity == CleavageSpecificity.None)
             {
-                if(specificProtease.Name != "top-down biomarker")
+                if (!topDownBiomarkerSearch)//standard top-down
                 {
                     // retain methionine
                     if ((initiatorMethionineBehavior != InitiatorMethionineBehavior.Cleave || protein[0] != 'M')
@@ -128,18 +127,14 @@ internal List<ProteolyticPeptide> GetUnmodifiedPeptides(Protein protein, int max
                         peptides.Add(new ProteolyticPeptide(protein, 2, protein.Length, 0, CleavageSpecificity.Full, "full:M cleaved"));
                     }
                 }
-                else
-                {
-                    protein.AddBiomarkers(true, true, true, true, initiatorMethionineBehavior, minPeptideLength, 7);
-                }
 
                 // Also digest using the proteolysis product start/end indices
                 peptides.AddRange(
                     protein.ProteolysisProducts
                     .Where(proteolysisProduct => proteolysisProduct.OneBasedEndPosition.HasValue && proteolysisProduct.OneBasedBeginPosition.HasValue
                         && OkayLength(proteolysisProduct.OneBasedEndPosition.Value - proteolysisProduct.OneBasedBeginPosition.Value + 1, minPeptideLength, maxPeptideLength))
                     .Select(proteolysisProduct =>
-                        new ProteolyticPeptide(protein, proteolysisProduct.OneBasedBeginPosition.Value, proteolysisProduct.OneBasedEndPosition.Value, 0, CleavageSpecificity.Full, proteolysisProduct.Type)));
+                        new ProteolyticPeptide(protein, proteolysisProduct.OneBasedBeginPosition.Value, proteolysisProduct.OneBasedEndPosition.Value, 0, CleavageSpecificity.None, proteolysisProduct.Type)));
             }
 
             // Full proteolytic cleavage
@@ -548,7 +543,6 @@ private static IEnumerable<ProteolyticPeptide> FixedTermini(int nTerminusProtein
             return intervals.Concat(fixedCTermIntervals).Concat(fixedNTermIntervals);
         }
 
-
         /// <summary>
         /// Gets peptides for the singleN protease
         /// </summary>

mzLib/Proteomics/ProteolyticDigestion/ProteinDigestion.cs

@@ -232,9 +232,9 @@ public IEnumerable<ProteolyticPeptide> SpeedySemiSpecificDigestion(Protein prote
         /// </summary>
         /// <param name="protein"></param>
         /// <returns></returns>
-        public IEnumerable<ProteolyticPeptide> Digestion(Protein protein)
+        public IEnumerable<ProteolyticPeptide> Digestion(Protein protein, bool topDownBiomarkerSearch = false)
         {
-            return Protease.GetUnmodifiedPeptides(protein, MaximumMissedCleavages, InitiatorMethionineBehavior, MinPeptideLength, MaxPeptideLength, DigestionParams.SpecificProtease);
+            return Protease.GetUnmodifiedPeptides(protein, MaximumMissedCleavages, InitiatorMethionineBehavior, MinPeptideLength, MaxPeptideLength, DigestionParams.SpecificProtease, topDownBiomarkerSearch);
         }
     }
 }

mzLib/Test/TestClassExtensions.cs

@@ -0,0 +1,25 @@
+using NUnit.Framework;
+using System.Diagnostics.CodeAnalysis;
+using MzLibUtil;
+using System.IO;
+using System.Linq;
+using System;
+
+namespace Test
+{
+    [TestFixture]
+    [ExcludeFromCodeCoverage]
+    public class TestClassExtensions
+    {
+        [Test]
+        public static void TestBoxCarSmooth()
+        {
+            double[] inputData = new double[] { 0.19, 0.69, 0.03, 0.85, 0.84, 0.46, 0.09, 0.05, 0.11, 0.5, 0.6, 0.78, 0.48, 0.66, 0.61, 0.78, 0.82, 0.18, 0.77, 0.14, 0.97, 0.48, 0.54, 0.98, 0.01, 0.38, 0.26, 0.4, 0.31, 0.41, 0.03, 0.2, 0.98, 0.36, 0.24, 0.51, 0.14, 0.96, 0.32, 0.9, 0.36, 0.57, 0.97, 0.07, 0.12, 0.73, 0.92, 0.51, 0.04, 0.2, 0.39, 0.32, 0.33, 0.62, 0.32, 0.68, 0.91, 0.3, 0.68, 0.22, 0.89, 0.27, 0.68, 0.08, 0.61, 0.25, 0.82, 0.73, 0.49, 0.76, 0.01, 0.15, 0.13, 0.96, 0.57, 0.58, 0.96, 0.93, 0.5, 0.45, 0.89, 0.44, 0.59, 0.68, 0.71, 0.85, 0.16, 0.18, 0.68, 0.37, 0.22, 0.81, 0.53, 0.26, 0.94, 0.52, 0.66, 0.55, 0.51, 0.14 };
+            double[] mySmoothedArray = ClassExtensions.BoxCarSmooth(inputData, 3);
+            string[] expectedOutput = new string[] { "0.3", "0.52", "0.57", "0.72", "0.46", "0.2", "0.08", "0.22", "0.4", "0.63", "0.62", "0.64", "0.58", "0.68", "0.74", "0.59", "0.59", "0.36", "0.63", "0.53", "0.66", "0.67", "0.51", "0.46", "0.22", "0.35", "0.32", "0.37", "0.25", "0.21", "0.4", "0.51", "0.53", "0.37", "0.3", "0.54", "0.47", "0.73", "0.53", "0.61", "0.63", "0.54", "0.39", "0.31", "0.59", "0.72", "0.49", "0.25", "0.21", "0.3", "0.35", "0.42", "0.42", "0.54", "0.64", "0.63", "0.63", "0.4", "0.6", "0.46", "0.61", "0.34", "0.46", "0.31", "0.56", "0.6", "0.68", "0.66", "0.42", "0.31", "0.1", "0.41", "0.55", "0.7", "0.7", "0.82", "0.8", "0.63", "0.61", "0.59", "0.64", "0.57", "0.66", "0.75", "0.57", "0.4", "0.34", "0.41", "0.42", "0.47", "0.52", "0.53", "0.58", "0.57", "0.71", "0.58", "0.57", "0.4" };
+            string[] actualOutput = mySmoothedArray.Select(v=>Math.Round(v,2).ToString()).ToArray();
+
+            CollectionAssert.AreEquivalent(expectedOutput, actualOutput);
+        }
+    }
+}