bug fix Modification Info List which reports the the modification stoichiometry in PSM counts and fractions on the full protein sequence

MetaMorpheus/EngineLayer/ProteinParsimony/ProteinGroup.cs

@@ -402,19 +402,23 @@ public void CalculateSequenceCoverage()
                 if (psm.BaseSequence != null)
                 {
                     psm.GetAminoAcidCoverage();
-                    var peptides = psm.BestMatchingPeptides.Select(p => p.Peptide);
-                    foreach (var peptide in peptides)
+                    var peptides = psm.BestMatchingPeptides.Select(p => p.Peptide).ToList();
+                    peptides = peptides.DistinctBy(p => p.FullSequence).ToList();
+
+                    if (peptides.Any())
                     {
-                        // might be unambiguous but also shared; make sure this protein group contains this peptide+protein combo
-                        if (Proteins.Contains(peptide.Protein))
+                        foreach (var peptide in peptides)
                         {
-                            proteinsWithUnambigSeqPsms[peptide.Protein].Add(peptide);
-                            //proteinsWithUnambigSeqPsmsCoverage[peptide.Protein].Add((peptide, psm.FragmentCoveragePositionInPeptide));
-
-                            // null FullSequence means that mods were not successfully localized; do not display them on the sequence coverage mods info
-                            if (psm.FullSequence != null)
+                            // might be unambiguous but also shared; make sure this protein group contains this peptide+protein combo
+                            if (Proteins.Contains(peptide.Protein))
                             {
-                                proteinsWithPsmsWithLocalizedMods[peptide.Protein].Add(peptide);
+                                proteinsWithUnambigSeqPsms[peptide.Protein].Add(peptide);
+
+                                // null FullSequence means that mods were not successfully localized; do not display them on the sequence coverage mods info
+                                if (peptide.FullSequence != null)
+                                {
+                                    proteinsWithPsmsWithLocalizedMods[peptide.Protein].Add(peptide);
+                                }
                             }
                         }
                     }

MetaMorpheus/EngineLayer/ProteolyticDigestion/proteases.tsv

@@ -1,23 +1,24 @@
 Name	Sequences Inducing Cleavage	Sequences Preventing Cleavage	Cleavage Terminus	Cleavage Specificity	PSI-MS Accession Number	PSI-MS Name	Site Regular Expression	Cleavage Mass Shifts	Notes
 Arg-C	R|			full	MS:1001303	Arg-C	(?<=R)(?!P)		
 Asp-N	|D			full	MS:1001304	Asp-N	(?=[BD])		
-chymotrypsin (don't cleave before proline)	"F[P]|,W[P]|,Y[P]|"			full	MS:1001306	Chymotrypsin	(?<=[FYWL])(?!P)		
-chymotrypsin (cleave before proline)	"F|,W|,Y|"			full	MS:1001306	Chymotrypsin	(?<=[FYWL])		
-CNBr	M|			full	MS:1001307	CNBr	(?<=M)	Homoserine lactone on M
-elastase	"A|,V|,S|,G|,L|,I|"			full		Elastase	(?<=[AVSGLI])		
+chymotrypsin (don't cleave before proline)	F[P]|,W[P]|,Y[P]|			full	MS:1001306	Chymotrypsin	(?<=[FYWL])(?!P)		
+chymotrypsin (cleave before proline)	F|,W|,Y|			full	MS:1001306	Chymotrypsin	(?<=[FYWL])		
+CNBr	M|			full	MS:1001307	CNBr	(?<=M)	Homoserine lactone on M	
+elastase(don't cleave before proline 15mc)	I[P]|,S[P]|,L[P]|,V[P]|,N[P]|,T[P]|,K[P]|,A[P]|,G[P]|,Y[P]|,Q[P]|,R[P]|,F[P]|,D[P]|			full		Elastase/P	(?<=[IPSPLPVPNPTPKPAPGPYPQPRPFD])(?!P)		
 Glu-C	E|			full					
-Glu-C (with asp)	"E|,D|"			full					
+Glu-C (with asp)	E|,D|			full					
 Lys-C (don't cleave before proline)	K[P]|			full	MS:1001309	Lys-C	(?<=K)(?!P)		
 Lys-C (cleave before proline)	K|			full	MS:1001310	Lys-C/P	(?<=K)		
 Lys-N	|K			full					
-semi-trypsin	"K|,R|"			semi	MS:1001313	Trypsin/P	(?<=[KR])		
-trypsin	"K|,R|"			full	MS:1001313	Trypsin/P	(?<=[KR])		
+semi-trypsin	K|,R|			semi	MS:1001313	Trypsin/P	(?<=[KR])		
+trypsin	K|,R|			full	MS:1001313	Trypsin/P	(?<=[KR])		
 tryptophan oxidation	W|			full					
 non-specific	X|			full	MS:1001956	unspecific cleavage			
-top-down				none	MS:1001955	no cleavage						
+top-down				none	MS:1001955	no cleavage			
 singleN				SingleN	MS:1001957	single cleavage			
 singleC				SingleC	MS:1001958	single cleavage			
 peptidomics				none		no cleavage			
 collagenase	GPX|GPX			full					
-StcE-trypsin	"TX|T,TX|S,SX|T,SX|S,K|,R|"			full		StcE/Trpsin			
-ProAlanase	"P|,A|"			full					
+StcE-trypsin	TX|T,TX|S,SX|T,SX|S,K|,R|			full		StcE/Trpsin			
+ProAlanase	P|,A|			full					
+subtilisin(don't cleave before proline 15mc)	N[P]|,S[P]|,L[P]|,K[P]|,I[P]|,D[P]|,Y[P]|,V[P]|,G[P]|,F[P]|,T[P]|,E[P]|,Q[P]|,A[P]|,R[P]|			full		Subtilisin/P	(?<=[NSLKIDYVGFTEQAR])(?!P)		

MetaMorpheus/Test/ProteinGroupTest.cs

@@ -2,11 +2,16 @@
 using NUnit.Framework;
 using Proteomics;
 using System.Collections.Generic;
+using System.Linq;
 using Proteomics.ProteolyticDigestion;
 using MassSpectrometry;
 using Chemistry;
+using EngineLayer.ClassicSearch;
 using FlashLFQ;
+using TaskLayer;
 using ProteinGroup = EngineLayer.ProteinGroup;
+using System.IO;
+using UsefulProteomicsDatabases;
 
 namespace Test
 {
@@ -163,6 +168,62 @@ public static void ProteinGroupDisplayModsTestWithGetIdentifiedPeptidesMethod()
             //This test just gets some lines in ProteinGroup covered. There is no accessible way to get the output of this method.
             Assert.DoesNotThrow(()=>proteinGroup1.GetIdentifiedPeptidesOutput(new List<SilacLabel>()));
         }
+
+        [Test]
+        public static void TestModificationInfoListInProteinGroupsOutput()
+        {
+            //Create GPTMD Task
+            //Create Search Task
+            GptmdTask task1 = new GptmdTask
+            {
+                CommonParameters = new CommonParameters(),
+                GptmdParameters = new GptmdParameters
+                {
+                    ListOfModsGptmd = GlobalVariables.AllModsKnown.Where(b =>
+                        b.ModificationType.Equals("Common Artifact")
+                        || b.ModificationType.Equals("Common Biological")
+                        || b.ModificationType.Equals("Metal")
+                        || b.ModificationType.Equals("Less Common")
+                        ).Select(b => (b.ModificationType, b.IdWithMotif)).ToList()
+                }
+            };
+
+            SearchTask task2 = new SearchTask
+            {
+                CommonParameters = new CommonParameters(),
+
+                SearchParameters = new SearchParameters
+                {
+                    DoParsimony = true,
+                    SearchTarget = true,
+                    WritePrunedDatabase = true,
+                    SearchType = SearchType.Classic
+                }
+            };
+            List<(string, MetaMorpheusTask)> taskList = new List<(string, MetaMorpheusTask)> { ("task1", task1), ("task2", task2) };
+            string mzmlName = @"TestData\PrunedDbSpectra.mzml";
+            string fastaName = @"TestData\DbForPrunedDb.fasta";
+            string outputFolder = Path.Combine(TestContext.CurrentContext.TestDirectory, @"TestPrunedGeneration");
+            var engine = new EverythingRunnerEngine(taskList, new List<string> { mzmlName }, new List<DbForTask> { new DbForTask(fastaName, false) }, outputFolder);
+            engine.Run();
+            string final = Path.Combine(MySetUpClass.outputFolder, "task2", "DbForPrunedDbGPTMDproteinPruned.xml");
+            List<Protein> proteins = ProteinDbLoader.LoadProteinXML(final, true, DecoyType.Reverse, new List<Modification>(), false, new List<string>(), out var ok);
+            // ensures that protein out put contains the correct number of proteins to match the following conditions.
+            // all proteins in DB have baseSequence!=null (not ambiguous)
+            // all proteins that belong to a protein group are written to DB
+            Assert.AreEqual(18, proteins.Count);
+            int totalNumberOfMods = proteins.Sum(p => p.OneBasedPossibleLocalizedModifications.Count + p.SequenceVariations.Sum(sv => sv.OneBasedModifications.Count));
+
+            //tests that modifications are being done correctly
+            Assert.AreEqual(0, totalNumberOfMods);
+
+            List<string> proteinGroupsOutput = File.ReadAllLines(Path.Combine(outputFolder, "task2", "AllQuantifiedProteinGroups.tsv")).ToList();
+            string firstDataLine = proteinGroupsOutput[2];
+            string modInfoListProteinTwo = firstDataLine.Split('\t')[14];
+            Assert.AreEqual("#aa66[Hydroxylation on K,info:occupancy=0.33(1/3)];#aa71[Oxidation on S,info:occupancy=0.67(2/3)]", modInfoListProteinTwo);
+
+            Directory.Delete(outputFolder, true);
+        }
     }
 }