restore peptide index support (#71)

* add CometPeptideIndex files * use proper mutex to protect g_pvDBIndex and fix capturing proper protein file position for peptide index search * update VS project with CometPeptideIndex.cpp and .h files * Add CreatePeptideIndex and CreateFragmentIndex to CometWrapper. * Get peptide index working for DoSingleSpectrumSearchMultiResults call thru CometWrapper. * address memory leak in GetPrevNextAA that showed up with peptide index RTS * Skip some peptide index parsing on subsequent calls to DoSingleSpectrumSearchMultiResults().
UWPR · Jan 2, 2025 · 292818c · 292818c
1 parent c884d4e
commit 292818c
Show file tree

Hide file tree

Showing 30 changed files with 1,739 additions and 366 deletions.
diff --git a/Comet.cpp b/Comet.cpp
@@ -100,7 +100,8 @@ void Usage(char *pszCmd)
    logout("                 -F<num>    to specify the first/start scan to search, overriding entry in parameters file\n");
    logout("                 -L<num>    to specify the last/end scan to search, overriding entry in parameters file\n");
    logout("                            (-L option is required if -F option is used)\n");
-   logout("                 -i         create peptide index file only (specify .idx file as database for index search)\n");
+   logout("                 -i         create .idx file for fragment ion indexing\n");
+   logout("                 -j         create .idx file for peptide indexing\n");
    logout("\n");
    sprintf(szTmp, "       example:  %s file1.mzXML file2.mzXML\n", pszCmd);
    logout(szTmp);
@@ -197,7 +198,15 @@ void SetOptions(char *arg,
          break;
       case 'i':
          sprintf(szParamStringVal, "1");
-         pSearchMgr->SetParam("create_index", szParamStringVal, 1);
+         pSearchMgr->SetParam("create_fragment_index", szParamStringVal, 1);
+         sprintf(szParamStringVal, "0");
+         pSearchMgr->SetParam("create_peptide_index", szParamStringVal, 0);
+         break;
+      case 'j':
+         sprintf(szParamStringVal, "0");
+         pSearchMgr->SetParam("create_fragment_index", szParamStringVal, 0);
+         sprintf(szParamStringVal, "1");
+         pSearchMgr->SetParam("create_peptide_index", szParamStringVal, 1);
          break;
       default:
          break;

diff --git a/CometSearch/CometCheckForUpdates.h b/CometSearch/CometCheckForUpdates.h
@@ -17,7 +17,7 @@
 #define _COMETCHECKFORUPDATES_H_
 
 #include "Common.h"
-#include "CometDataInternal.h"
+//#include "CometDataInternal.h"
 
 #include <errno.h>
 #include <string.h>

diff --git a/CometSearch/CometDataInternal.h b/CometSearch/CometDataInternal.h
@@ -134,7 +134,8 @@ struct Options
    int bSkipAlreadyDone;         // 0=search everything; 1=don't re-search if .out exists
    int bMango;                   // 0=normal; 1=Mango x-link ms2 input
    int bScaleFragmentNL;         // 0=no; 1=scale fragment NL for each modified residue contained in fragment
-   int bCreateIndex;             // 0=normal search; 1=create peptide index file
+   int bCreateFragmentIndex;     // 0=normal search; 1=create fragment ion index file
+   int bCreatePeptideIndex;      // 0=normal search; 1=create peptide index file; only one of bCreateFragmentIndex and bCreatePeptideIndex can be 1
    int bVerboseOutput;
    int bShowFragmentIons;
    int bExplicitDeltaCn;         // if set to 1, do not use sequence similarity logic
@@ -194,7 +195,8 @@ struct Options
       bSkipAlreadyDone = a.bSkipAlreadyDone;
       bMango = a.bMango;
       bScaleFragmentNL = a.bScaleFragmentNL;
-      bCreateIndex = a.bCreateIndex;
+      bCreatePeptideIndex = a.bCreatePeptideIndex;
+      bCreateFragmentIndex = a.bCreateFragmentIndex;
       bVerboseOutput = a.bVerboseOutput;
       bShowFragmentIons = a.bShowFragmentIons;
       bExplicitDeltaCn = a.bExplicitDeltaCn;
@@ -712,12 +714,13 @@ struct StaticParams
    double          dOneMinusBinOffset;  // this is used in BIN() many times so calculate once
    IonInfo         ionInformation;
    int             iXcorrProcessingOffset;
-   int             bIndexDb;            // 0 = normal fasta; 1 = indexed database
+   int             iIndexDb;            // 0 = normal fasta; 1 = fragment ion indexed; 2 = peptide index
    vector<double>  vectorMassOffsets;
    vector<double>  precursorNLIons;
    int             iPrecursorNLSize;
    int             iOldModsEncoding;
    bool            bSkipToStartScan;
+   std::chrono::high_resolution_clock::time_point tRealTimeStart;     // track run time of real-time index search
 
    StaticParams()
    {
@@ -767,7 +770,7 @@ struct StaticParams
       szMod[0] = '\0';
 
       iXcorrProcessingOffset = 75;
-      bIndexDb = 0;
+      iIndexDb = 0;
 
       databaseInfo.szDatabase[0] = '\0';
 
@@ -882,7 +885,8 @@ struct StaticParams
       options.bSkipAlreadyDone = 1;
       options.bMango = 0;
       options.bScaleFragmentNL = 0;
-      options.bCreateIndex = 0;
+      options.bCreatePeptideIndex = 0;
+      options.bCreateFragmentIndex = 0;
       options.bVerboseOutput = 0;
       options.iDecoySearch = 0;
       options.iNumThreads = 4;
@@ -949,7 +953,7 @@ extern StaticParams    g_staticParams;
 
 extern string g_psGITHUB_SHA;             // grab the GITHUB_SHA environment variable and trim to 7 chars; null if environment variable not present
 
-extern vector<DBIndex> g_pvDBIndex;
+extern vector<DBIndex> g_pvDBIndex;       // used in both peptide index and fragment ion index; latter to store plain peptides
 
 extern vector<vector<comet_fileoffset_t>> g_pvProteinsList;
 
@@ -972,6 +976,8 @@ extern int* PEPTIDE_MOD_SEQ_IDXS;
 
 extern int MOD_NUM;
 extern bool g_bPlainPeptideIndexRead;   // set to true if plain peptide index file is read (and fragment index generated)
+                                        // poor choice of name for the fragment index .idx given peptide index is back
+extern bool g_bPeptideIndexRead;        // set to true if peptide index file is read
 
 // Query stores information for peptide scoring and results
 // This struct is allocated for each spectrum/charge combination

diff --git a/CometSearch/CometFragmentIndex.cpp b/CometSearch/CometFragmentIndex.cpp
@@ -13,12 +13,10 @@
 // limitations under the License.
 
 
-#include "Common.h"
 #include "CometFragmentIndex.h"
 #include "CometSearch.h"
 #include "ThreadPool.h"
 #include "CometStatus.h"
-//#include "CometPostAnalysis.h"
 #include "CometMassSpecUtils.h"
 #include "ModificationsPermuter.h"
 
@@ -37,7 +35,7 @@ int MOD_NUM = 0;
 
 Mutex CometFragmentIndex::_vFragmentPeptidesMutex;
 
-//comet_fileoffset_t clSizeCometFileOffset;
+
 #ifdef _WIN32
 #ifdef _WIN64
 comet_fileoffset_t clSizeCometFileOffset = sizeof(comet_fileoffset_t);              //win64
@@ -48,6 +46,7 @@ comet_fileoffset_t clSizeCometFileOffset = (long long)sizeof(comet_fileoffset_t)
 comet_fileoffset_t clSizeCometFileOffset = sizeof(comet_fileoffset_t);              //linux
 #endif
 
+
 CometFragmentIndex::CometFragmentIndex()
 {
 }
@@ -652,7 +651,7 @@ bool CometFragmentIndex::WritePlainPeptideIndex(ThreadPool *tp)
       exit(1);
    }
 
-   strOut = " Creating plain peptide/protein index file:\n";
+   strOut = " Creating plain peptide/protein index file for fragment ion indexing:\n";
    logout(strOut.c_str());
    fflush(stdout);
    strOut = " - parse peptides from database ... ";
@@ -671,15 +670,15 @@ bool CometFragmentIndex::WritePlainPeptideIndex(ThreadPool *tp)
 
    if (bSucceeded)
    {
-      g_staticParams.options.bCreateIndex = true;
-      g_staticParams.bIndexDb = false;
+      g_staticParams.options.bCreateFragmentIndex = true;
+      g_staticParams.iIndexDb = 0;
 
       // this step calls RunSearch just to pull out all peptides
       // to write into the .idx pepties/proteins file
       bSucceeded = CometSearch::RunSearch(0, 0, tp);
 
-      g_staticParams.options.bCreateIndex = false;
-      g_staticParams.bIndexDb = true;
+      g_staticParams.options.bCreateFragmentIndex = false;
+      g_staticParams.iIndexDb = 1;
    }
 
    if (bSwapIdxExtension)
@@ -767,7 +766,7 @@ bool CometFragmentIndex::WritePlainPeptideIndex(ThreadPool *tp)
    cout << " - write peptides/proteins to file" << endl;
 
    // write out index header
-   fprintf(fp, "Comet peptide index.  Comet version %s\n", g_sCometVersion.c_str());
+   fprintf(fp, "Comet fragment ion index plain peptides.  Comet version %s\n", g_sCometVersion.c_str());
    fprintf(fp, "InputDB:  %s\n", g_staticParams.databaseInfo.szDatabase);
    fprintf(fp, "MassRange: %lf %lf\n", g_staticParams.options.dPeptideMassLow, g_staticParams.options.dPeptideMassHigh);
    fprintf(fp, "LengthRange: %d %d\n", g_staticParams.options.peptideLengthRange.iStart, g_staticParams.options.peptideLengthRange.iEnd);
@@ -893,7 +892,7 @@ bool CometFragmentIndex::ReadPlainPeptideIndex(void)
    if (g_bPlainPeptideIndexRead)
       return 1;
 
-   if (g_staticParams.options.bCreateIndex && !strstr(g_staticParams.databaseInfo.szDatabase + strlen(g_staticParams.databaseInfo.szDatabase) - 4, ".idx"))
+   if (g_staticParams.options.bCreateFragmentIndex && !strstr(g_staticParams.databaseInfo.szDatabase + strlen(g_staticParams.databaseInfo.szDatabase) - 4, ".idx"))
       strIndexFile = g_staticParams.databaseInfo.szDatabase + string(".idx");
    else // database already is .idx
       strIndexFile = g_staticParams.databaseInfo.szDatabase;

diff --git a/CometSearch/CometFragmentIndex.h b/CometSearch/CometFragmentIndex.h
@@ -17,7 +17,6 @@
 #define _COMETFRAGMENTINDEX_H_
 
 #include "Common.h"
-#include "CometDataInternal.h"
 #include "CometSearch.h"
 #include <functional>
 
@@ -32,6 +31,10 @@ class CometFragmentIndex
    static bool CreateFragmentIndex(ThreadPool *tp);
    static string ElapsedTime(std::chrono::time_point<std::chrono::steady_clock> tStartTime);
    static int WhichPrecursorBin(double dMass);
+   static bool CompareByPeptide(const DBIndex &lhs,
+                                const DBIndex &rhs);
+   static bool CompareByMass(const DBIndex &lhs,
+                             const DBIndex &rhs);
 
 private:
 
@@ -52,16 +55,12 @@ class CometFragmentIndex
                                       unsigned int y);
    static void SortFragmentThreadProc(int iWhichThread,
                                       ThreadPool* tp);
-   static bool CompareByPeptide(const DBIndex &lhs,
-                                const DBIndex &rhs);
-   static bool CompareByMass(const DBIndex &lhs,
-                             const DBIndex &rhs);
-
+/*
    unsigned int       _uiBinnedIonMasses[MAX_FRAGMENT_CHARGE + 1][NUM_ION_SERIES][MAX_PEPTIDE_LEN][VMODS + 1];
    unsigned int       _uiBinnedIonMassesDecoy[MAX_FRAGMENT_CHARGE + 1][NUM_ION_SERIES][MAX_PEPTIDE_LEN][VMODS + 1];
    unsigned int       _uiBinnedPrecursorNL[MAX_PRECURSOR_NL_SIZE][MAX_PRECURSOR_CHARGE];
    unsigned int       _uiBinnedPrecursorNLDecoy[MAX_PRECURSOR_NL_SIZE][MAX_PRECURSOR_CHARGE];
-
+*/
    static bool *_pbSearchMemoryPool;    // Pool of memory to be shared by search threads
    static bool **_ppbDuplFragmentArr;   // Number of arrays equals number of threads
 

diff --git a/CometSearch/CometInterfaces.h b/CometSearch/CometInterfaces.h
@@ -29,7 +29,8 @@ namespace CometInterfaces
    {
 public:
       virtual ~ICometSearchManager() {}
-      virtual bool CreateIndex() = 0;
+      virtual bool CreateFragmentIndex() = 0;
+      virtual bool CreatePeptideIndex() = 0;
       virtual bool DoSearch() = 0;
       virtual bool InitializeSingleSpectrumSearch() = 0;
       virtual void FinalizeSingleSpectrumSearch() = 0;
@@ -43,15 +44,15 @@ namespace CometInterfaces
                                           vector<Fragment> & matchedFragments,
                                           Scores & scores) = 0;
       virtual bool DoSingleSpectrumSearchMultiResults(const int topN,
-          const int iPrecursorCharge,
-          const double dMZ,
-          double* dMass,
-          double* dInten,
-          const int iNumPeaks,
-          vector<string>& strReturnPeptide,
-          vector<string>& strReturnProtein,
-          vector<vector<Fragment>>& matchedFragments,
-          vector<Scores>& scores) = 0;
+                                                      const int iPrecursorCharge,
+                                                      const double dMZ,
+                                                      double* dMass,
+                                                      double* dInten,
+                                                      const int iNumPeaks,
+                                                      vector<string>& strReturnPeptide,
+                                                      vector<string>& strReturnProtein,
+                                                      vector<vector<Fragment>>& matchedFragments,
+                                                      vector<Scores>& scores) = 0;
       virtual void AddInputFiles(vector<InputFileInfo*> &pvInputFiles) = 0;
       virtual void SetOutputFileBaseName(const char *pszBaseName) = 0;
       virtual void SetParam(const string &name, const string &strValue, const string &value) = 0;