Remove code for ignoring ambiguous spectra from MSFReader.cpp (#3111)

Fixed "Keep ambiguous matches" not working for .pdResult files (reported by Wes)
ProteoWizard · Jan 17, 2025 · abe9cec · abe9cec
1 parent 9b90309
commit abe9cec
Show file tree

Hide file tree

Showing 5 changed files with 64 additions and 96 deletions.
diff --git a/pwiz_tools/BiblioSpec/src/MSFReader.cpp b/pwiz_tools/BiblioSpec/src/MSFReader.cpp
@@ -310,37 +310,6 @@ namespace BiblioSpec
 
     void MSFReader::collectPsms() {
         sqlite3_stmt* statement;
-        map<int, double> alts; // peptide id --> alt score, for breaking ties when q-values are identical
-        vector<string> altScoreNames;
-        altScoreNames.push_back("XCorr");
-        altScoreNames.push_back("IonScore");
-
-        if (tableExists(msfFile_, "TargetPsms")) {
-            for (vector<string>::const_iterator i = altScoreNames.begin(); i != altScoreNames.end(); ++i) {
-                if (!columnExists(msfFile_, "TargetPsms", *i)) {
-                    continue;
-                }
-                statement = getStmt("SELECT PeptideID, " + *i + " FROM TargetPsms");
-                while (hasNext(&statement)) {
-                    alts[sqlite3_column_int(statement, 0)] = sqlite3_column_double(statement, 1);
-                }
-                break;
-            }
-        } else if (tableExists(msfFile_, "PeptideScores") && tableExists(msfFile_, "ProcessingNodeScores")) {
-            for (vector<string>::const_iterator i = altScoreNames.begin(); i != altScoreNames.end(); ++i) {
-                statement = getStmt(
-                    "SELECT PeptideID, ScoreValue "
-                    "FROM PeptideScores JOIN ProcessingNodeScores ON PeptideScores.ScoreID = ProcessingNodeScores.ScoreID "
-                    "WHERE ScoreName = '" + *i + "'");
-                while (hasNext(&statement)) {
-                    alts[sqlite3_column_int(statement, 0)] = sqlite3_column_double(statement, 1);
-                }
-                if (!alts.empty()) {
-                    break;
-                }
-            }
-        }
-
         int resultCount, pepConfidence, protConfidence;
         PSM_SCORE_TYPE scoreType;
         getScoreInfo(&statement, &resultCount, &scoreType, &pepConfidence, &protConfidence);
@@ -349,7 +318,6 @@ namespace BiblioSpec
         ProgressIndicator progress(resultCount);
 
         initFileNameMap();
-        map<string, ProcessedMsfSpectrum> processedSpectra;
         ModSet modSet = ModSet(msfFile_, !versionLess(2, 2) || filtered_);
         map<int, int> fileIdMap = getFileIds();
 
@@ -390,56 +358,7 @@ namespace BiblioSpec
                 continue;
             }
 
-            auto altIter = alts.find(peptideId);    
-            double altScore = (altIter != alts.end()) ? altIter->second : -std::numeric_limits<double>::max();
-
-            // check if we already processed a peptide that references this spectrum
-            auto processedSpectraSearch = processedSpectra.find(specId);
-            if (processedSpectraSearch != processedSpectra.end()) {
-                ProcessedMsfSpectrum& processed = processedSpectraSearch->second;
-                // not an ambigous spectrum (yet)
-                if (!processed.ambiguous) {
-                    if (qvalue > processed.qvalue || (qvalue == processed.qvalue && altScore < processed.altScore)) { // worse than other score, skip this
-                        Verbosity::debug("Peptide %d (%s) had a worse score than another peptide (%s) "
-                                         "referencing spectrum %d (ignoring this peptide).",
-                                         peptideId, sequence.c_str(), processed.psm->unmodSeq.c_str(), specId.c_str());
-                        continue;
-                    } else if (qvalue == processed.qvalue && altScore == processed.altScore) { // equal, discard other and skip this
-                        Verbosity::debug("Peptide %d (%s) had the same score as another peptide (%s) "
-                                         "referencing spectrum %d (ignoring both peptides).",
-                                         peptideId, sequence.c_str(), processed.psm->unmodSeq.c_str(), specId.c_str());
-
-                        removeFromFileMap(processed.psm);
-                        delete processed.psm;
-
-                        processed.psm = NULL;
-                        processed.ambiguous = true;
-                        continue;
-                    } else { // better than other score, discard other
-                        Verbosity::debug("Peptide %d (%s) had a better score than another peptide (%s) "
-                                         "referencing spectrum %d (ignoring other peptide).",
-                                         peptideId, sequence.c_str(), processed.psm->unmodSeq.c_str(), specId.c_str());
-                        removeFromFileMap(processed.psm);
-                        curPSM_ = processed.psm;
-                        curPSM_->mods.clear();
-                        processed.qvalue = qvalue;
-                        processed.altScore = altScore;
-                    }
-                } else { // ambigous spectrum, check if score is better
-                    Verbosity::debug("Peptide %d (%s) with score %f references same spectrum as other peptides "
-                                     "that had score %f.", peptideId, sequence.c_str(), qvalue, processed.qvalue);
-                    if (qvalue < processed.qvalue || (qvalue == processed.qvalue && altScore > processed.altScore)) {
-                        curPSM_ = new PSM();
-                        processedSpectraSearch->second = ProcessedMsfSpectrum(curPSM_, qvalue, altScore);
-                    } else {
-                        continue;
-                    }
-                }
-            } else {
-                // unseen spectrum
-                curPSM_ = new PSM();
-                processedSpectra[specId] = ProcessedMsfSpectrum(curPSM_, qvalue, altScore);
-            }
+            curPSM_ = new PSM();
 
             if (findItr->second->charge > 0)
                 curPSM_->charge = findItr->second->charge;

diff --git a/pwiz_tools/BiblioSpec/tests/Jamfile.jam b/pwiz_tools/BiblioSpec/tests/Jamfile.jam
@@ -269,6 +269,7 @@ blib-test-build tinyPlink-proxl : -o : output/tinyPlink-proxl.blib : tinyPlink-p
 blib-test-build tinyPeptideProphet-proxl : -o : output/tinyPeptideProphet-proxl.blib : tinyPeptideProphet-proxl.check zbuild.skip-lines : $(TEST_INPUTS_PATH)/tinyPeptideProphet.proxl.xml ;
 blib-test-build tinyMerox-proxl : -o : output/tinyMerox-proxl.blib : tinyMerox-proxl.check zbuild.skip-lines : $(TEST_INPUTS_PATH)/tinyMerox.proxl.xml ;
 blib-test-build tiny-msf : --unicode -o : output/tiny-msf.blib : tiny-msf.check zbuild.skip-lines : $(TEST_INPUTS_PATH)/tiny.msf ;
+blib-test-build tiny-msf-keep : --unicode -o -K : output/tiny-msf-keep.blib : tiny-msf-keep.check zbuild.skip-lines : $(TEST_INPUTS_PATH)/tiny.msf ;
 blib-test-build tiny-v2-msf : -o : output/tiny-v2-msf.blib : tiny-v2-msf.check zbuild.skip-lines : $(TEST_INPUTS_PATH)/tiny-v2.msf ;
 blib-test-build tiny-v2-filtered-pdResult : -o : output/tiny-v2-filtered-pdResult.blib : tiny-v2-filtered-pdResult.check zbuild.skip-lines : $(TEST_INPUTS_PATH)/tiny-v2-filtered.pdResult ;
 blib-test-build md_special_filtered-pdResult : --unicode -o : output/md_special_filtered-pdResult.blib : md_special_filtered-pdResult.check zbuild.skip-lines : $(TEST_INPUTS_PATH)/md_special_filtered.pdResult ;

diff --git a/pwiz_tools/BiblioSpec/tests/reference/tiny-msf-keep.check b/pwiz_tools/BiblioSpec/tests/reference/tiny-msf-keep.check
@@ -0,0 +1,50 @@
+libLSID	numSpecs	majorVersion	minorVersion
+urn:lsid:proteome.gs.washington.edu:spectral_library:bibliospec:redundant:tiny-msf-keep.blib	6	1	10
+id	RefSpectraID	position	mass
+1	1	1	4.008491
+2	1	2	4.008491
+3	1	7	4.008491
+4	2	3	4.008491
+5	2	4	4.008491
+6	2	7	4.008491
+7	3	1	4.008491
+8	3	2	4.008491
+9	3	7	4.008491
+10	4	1	4.008491
+11	4	3	4.008491
+12	4	7	4.008491
+13	5	1	15.994915
+14	5	8	15.994915
+15	5	10	4.008491
+id	peptideSeq	precursorMZ	precursorCharge	peptideModSeq	prevAA	nextAA	copies	numPeaks	ionMobility	collisionalCrossSectionSqA	ionMobilityHighEnergyOffset	ionMobilityType	retentionTime	startTime	endTime	totalIonCurrent	moleculeName	chemicalFormula	precursorAdduct	inchiKey	otherKeys	fileID	SpecIDinFile	score	scoreType
+1	KKLVPLK	419.81906737	2	K[+4.0]K[+4.0]LVPLK[+4.0]	-	-	1	90	0.0	0.0	0.0	0	0.32088	N/A	N/A	3448.02592301	N/A	N/A	N/A	N/A	N/A	1	2	0.002	1
+2	VPKKILK	419.81915892	2	VPK[+4.0]K[+4.0]ILK[+4.0]	-	-	1	95	0.0	0.0	0.0	0	1.12651166	N/A	N/A	4332.04377079	N/A	N/A	N/A	N/A	N/A	1	4	0.006	1
+3	KKLVPLK	419.81915892	2	K[+4.0]K[+4.0]LVPLK[+4.0]	-	-	1	95	0.0	0.0	0.0	0	1.12651166	N/A	N/A	4332.04377079	N/A	N/A	N/A	N/A	N/A	1	4	0.009	1
+4	KLKVIPK	419.81921996	2	K[+4.0]LK[+4.0]VIPK[+4.0]	-	-	1	100	0.0	0.0	0.0	0	2.24853	N/A	N/A	4803.68319678	N/A	N/A	N/A	N/A	N/A	1	6	0.002	1
+5	MASASGAMAK	481.22015990	2	M[+16.0]ASASGAM[+16.0]AK[+4.0]	-	-	1	120	0.0	0.0	0.0	0	2.85028666	N/A	N/A	7404.39071369	N/A	N/A	N/A	N/A	N/A	1	9	0.01	1
+6	RPVTPKK	413.76981200	2	RPVTPKK	-	-	1	175	0.0	0.0	0.0	0	2.87913166	N/A	N/A	16087.57	N/A	N/A	N/A	N/A	N/A	2	10	0.002	1
+id	fileName	idFileName	cutoffScore
+1	D:/Elite_MAM/MITO_03.raw	/BiblioSpec/tests/inputs/试验_tiny.msf	-1.0
+2	X:/Test.raw	/BiblioSpec/tests/inputs/试验_tiny.msf	-1.0
+id	scoreType	probabilityType
+0	UNKNOWN	NOT_A_PROBABILITY_VALUE
+1	PERCOLATOR QVALUE	PROBABILITY_THAT_IDENTIFICATION_IS_INCORRECT
+2	PEPTIDE PROPHET SOMETHING	PROBABILITY_THAT_IDENTIFICATION_IS_CORRECT
+3	SPECTRUM MILL	NOT_A_PROBABILITY_VALUE
+4	IDPICKER FDR	PROBABILITY_THAT_IDENTIFICATION_IS_INCORRECT
+5	MASCOT IONS SCORE	PROBABILITY_THAT_IDENTIFICATION_IS_INCORRECT
+6	TANDEM EXPECTATION VALUE	PROBABILITY_THAT_IDENTIFICATION_IS_INCORRECT
+7	PROTEIN PILOT CONFIDENCE	PROBABILITY_THAT_IDENTIFICATION_IS_CORRECT
+8	SCAFFOLD SOMETHING	PROBABILITY_THAT_IDENTIFICATION_IS_CORRECT
+9	WATERS MSE PEPTIDE SCORE	NOT_A_PROBABILITY_VALUE
+10	OMSSA EXPECTATION SCORE	PROBABILITY_THAT_IDENTIFICATION_IS_INCORRECT
+11	PROTEIN PROSPECTOR EXPECTATION SCORE	PROBABILITY_THAT_IDENTIFICATION_IS_INCORRECT
+12	SEQUEST XCORR	PROBABILITY_THAT_IDENTIFICATION_IS_INCORRECT
+13	MAXQUANT SCORE	PROBABILITY_THAT_IDENTIFICATION_IS_INCORRECT
+14	MORPHEUS SCORE	PROBABILITY_THAT_IDENTIFICATION_IS_INCORRECT
+15	MSGF+ SCORE	PROBABILITY_THAT_IDENTIFICATION_IS_INCORRECT
+16	PEAKS CONFIDENCE SCORE	PROBABILITY_THAT_IDENTIFICATION_IS_INCORRECT
+17	BYONIC SCORE	PROBABILITY_THAT_IDENTIFICATION_IS_INCORRECT
+18	PEPTIDE SHAKER CONFIDENCE	PROBABILITY_THAT_IDENTIFICATION_IS_CORRECT
+19	GENERIC Q-VALUE	PROBABILITY_THAT_IDENTIFICATION_IS_INCORRECT
+20	HARDKLOR IDOTP	PROBABILITY_THAT_IDENTIFICATION_IS_CORRECT
diff --git a/pwiz_tools/BiblioSpec/tests/reference/tiny-msf.check b/pwiz_tools/BiblioSpec/tests/reference/tiny-msf.check
@@ -1,24 +1,20 @@
 libLSID	numSpecs	majorVersion	minorVersion
-urn:lsid:proteome.gs.washington.edu:spectral_library:bibliospec:redundant:tiny-msf.blib	5	1	10
+urn:lsid:proteome.gs.washington.edu:spectral_library:bibliospec:redundant:tiny-msf.blib	4	1	10
 id	RefSpectraID	position	mass
 1	1	1	4.008491
 2	1	2	4.008491
 3	1	7	4.008491
-4	2	3	4.008491
-5	2	4	4.008491
+4	2	1	4.008491
+5	2	3	4.008491
 6	2	7	4.008491
-7	3	1	4.008491
-8	3	3	4.008491
-9	3	7	4.008491
-10	4	1	15.994915
-11	4	8	15.994915
-12	4	10	4.008491
+7	3	1	15.994915
+8	3	8	15.994915
+9	3	10	4.008491
 id	peptideSeq	precursorMZ	precursorCharge	peptideModSeq	prevAA	nextAA	copies	numPeaks	ionMobility	collisionalCrossSectionSqA	ionMobilityHighEnergyOffset	ionMobilityType	retentionTime	startTime	endTime	totalIonCurrent	moleculeName	chemicalFormula	precursorAdduct	inchiKey	otherKeys	fileID	SpecIDinFile	score	scoreType
 1	KKLVPLK	419.81906737	2	K[+4.0]K[+4.0]LVPLK[+4.0]	-	-	1	90	0.0	0.0	0.0	0	0.32088	N/A	N/A	3448.02592301	N/A	N/A	N/A	N/A	N/A	1	2	0.002	1
-2	VPKKILK	419.81915892	2	VPK[+4.0]K[+4.0]ILK[+4.0]	-	-	1	95	0.0	0.0	0.0	0	1.12651166	N/A	N/A	4332.04377079	N/A	N/A	N/A	N/A	N/A	1	4	0.006	1
-3	KLKVIPK	419.81921996	2	K[+4.0]LK[+4.0]VIPK[+4.0]	-	-	1	100	0.0	0.0	0.0	0	2.24853	N/A	N/A	4803.68319678	N/A	N/A	N/A	N/A	N/A	1	6	0.002	1
-4	MASASGAMAK	481.22015990	2	M[+16.0]ASASGAM[+16.0]AK[+4.0]	-	-	1	120	0.0	0.0	0.0	0	2.85028666	N/A	N/A	7404.39071369	N/A	N/A	N/A	N/A	N/A	1	9	0.01	1
-5	RPVTPKK	413.76981200	2	RPVTPKK	-	-	1	175	0.0	0.0	0.0	0	2.87913166	N/A	N/A	16087.57	N/A	N/A	N/A	N/A	N/A	2	10	0.002	1
+2	KLKVIPK	419.81921996	2	K[+4.0]LK[+4.0]VIPK[+4.0]	-	-	1	100	0.0	0.0	0.0	0	2.24853	N/A	N/A	4803.68319678	N/A	N/A	N/A	N/A	N/A	1	6	0.002	1
+3	MASASGAMAK	481.22015990	2	M[+16.0]ASASGAM[+16.0]AK[+4.0]	-	-	1	120	0.0	0.0	0.0	0	2.85028666	N/A	N/A	7404.39071369	N/A	N/A	N/A	N/A	N/A	1	9	0.01	1
+4	RPVTPKK	413.76981200	2	RPVTPKK	-	-	1	175	0.0	0.0	0.0	0	2.87913166	N/A	N/A	16087.57	N/A	N/A	N/A	N/A	N/A	2	10	0.002	1
 id	fileName	idFileName	cutoffScore
 1	D:/Elite_MAM/MITO_03.raw	/BiblioSpec/tests/inputs/试验_tiny.msf	-1.0
 2	X:/Test.raw	/BiblioSpec/tests/inputs/试验_tiny.msf	-1.0
@@ -43,3 +39,4 @@ id	scoreType	probabilityType
 17	BYONIC SCORE	PROBABILITY_THAT_IDENTIFICATION_IS_INCORRECT
 18	PEPTIDE SHAKER CONFIDENCE	PROBABILITY_THAT_IDENTIFICATION_IS_CORRECT
 19	GENERIC Q-VALUE	PROBABILITY_THAT_IDENTIFICATION_IS_INCORRECT
+20	HARDKLOR IDOTP	PROBABILITY_THAT_IDENTIFICATION_IS_CORRECT
diff --git a/pwiz_tools/Skyline/TestPerf/PerfThermoFAIMSTest.cs b/pwiz_tools/Skyline/TestPerf/PerfThermoFAIMSTest.cs
@@ -120,7 +120,7 @@ private void TestPopulateDocumentFromLibrary()
             // Add all peptides
             var filterMatchedPeptidesDlg = ShowDialog<FilterMatchedPeptidesDlg>(viewLibUI.AddAllPeptides);
             var docBefore = WaitForProteinMetadataBackgroundLoaderCompletedUI();
-            using (new CheckDocumentState(1, 8433, 10882, 43484))
+            using (new CheckDocumentState(1, 8591, 11057, 44174))
             {
                 RunDlg<MultiButtonMsgDlg>(filterMatchedPeptidesDlg.OkDialog, addLibraryPepsDlg =>
                 {
@@ -146,6 +146,7 @@ private void TestWizardBuildDocumentLibraryAndFinish(string documentFile)
             RunUI(() =>
             {
                 Assert.IsTrue(importPeptideSearchDlg.CurrentPage == ImportPeptideSearchDlg.Pages.spectra_page);
+                importPeptideSearchDlg.BuildPepSearchLibControl.IncludeAmbiguousMatches = true;
                 importPeptideSearchDlg.BuildPepSearchLibControl.AddSearchFiles(SearchFiles);
             });
             WaitForConditionUI(() => importPeptideSearchDlg.IsEarlyFinishButtonEnabled);