Add (an overly) verbose mode for (maybe) helping to debug why bucketi…

…zer is failing. Issue marbl#959.
zhouxingchen1993 · Jun 19, 2018 · 3a2e695 · 3a2e695
1 parent d6cf139
commit 3a2e695
Show file tree

Hide file tree

Showing 4 changed files with 51 additions and 16 deletions.
diff --git a/src/stores/ovStore.H b/src/stores/ovStore.H
@@ -320,7 +320,7 @@ private:
 
 class ovStoreFilter {
 public:
-  ovStoreFilter(sqStore *seq_, double maxErate);
+  ovStoreFilter(sqStore *seq_, double maxErate, bool beVerbose = false);
   ~ovStoreFilter();
 
   void     filterOverlap(ovOverlap     &foverlap,
@@ -350,6 +350,8 @@ public:
   uint32   maxID;
   uint32   maxEvalue;
 
+  bool     beVerbose;
+
   uint64   saveUTG;
   uint64   saveOBT;
   uint64   saveDUP;

diff --git a/src/stores/ovStoreBucketizer.C b/src/stores/ovStoreBucketizer.C
@@ -66,6 +66,14 @@ writeToFile(sqStore          *seq,
     sliceSize[df] = 0;
   }
 
+  if ((df < 1) ||
+      (df > config->numSlices() + 1)) {
+    char ovlstr[256];
+
+    fprintf(stderr, "Invalid slice file %u in overlap %s\n",
+            df, overlap->toString(ovlstr, ovOverlapAsUnaligned, false));
+  }
+
   sliceFile[df]->writeOverlap(overlap);
   sliceSize[df]++;
 }
@@ -82,6 +90,7 @@ main(int argc, char **argv) {
   double          maxErrorRate   = 1.0;
 
   bool            forceOverwrite = false;
+  bool            beVerbose      = false;
 
   char            createName[FILENAME_MAX+1];
   char            sliceSName[FILENAME_MAX+1];
@@ -110,6 +119,9 @@ main(int argc, char **argv) {
     } else if (strcmp(argv[arg], "-f") == 0) {
       forceOverwrite = true;
 
+    } else if (strcmp(argv[arg], "-v") == 0) {
+      beVerbose = true;
+
     } else {
       char *s = new char [1024];
       snprintf(s, 1024, "%s: unknown option '%s'.\n", argv[0], argv[arg]);
@@ -141,6 +153,7 @@ main(int argc, char **argv) {
     fprintf(stderr, "  -e e                  filter overlaps above e fraction error\n");
     fprintf(stderr, "\n");
     fprintf(stderr, "  -f                    force overwriting existing data\n");
+    fprintf(stderr, "  -v                    be overly verbose\n");
     fprintf(stderr, "\n");
 
     for (uint32 ii=0; ii<err.size(); ii++)
@@ -180,9 +193,16 @@ main(int argc, char **argv) {
     exit(1);
   }
 
-  //  Report options.
+  //  Open inputs.
+
+  sqStore        *seq    = sqStore::sqStore_open(seqName);
 
   fprintf(stderr, "\n");
+  fprintf(stderr, "Opened '%s' with %u reads.\n", seqName, seq->sqStore_getNumReads());
+  fprintf(stderr, "\n");
+
+  //  Report options.
+
   fprintf(stderr, "Constructing slice " F_U32 " for store '%s'.\n", bucketNum, ovlName);
   fprintf(stderr, " - Filtering overlaps over %.4f fraction error.\n", maxErrorRate);
   fprintf(stderr, "\n");
@@ -192,10 +212,6 @@ main(int argc, char **argv) {
   AS_UTL_mkdir(ovlName);
   AS_UTL_mkdir(createName);
 
-  //  Open inputs.
-
-  sqStore        *seq    = sqStore::sqStore_open(seqName);
-
   //  Allocate stuff.
 
   ovFile        **sliceFile = new ovFile * [config->numSlices() + 1];
@@ -204,7 +220,7 @@ main(int argc, char **argv) {
   memset(sliceFile, 0, sizeof(ovFile *) * (config->numSlices() + 1));
   memset(sliceSize, 0, sizeof(uint64)   * (config->numSlices() + 1));
 
-  ovStoreFilter *filter = new ovStoreFilter(seq, maxErrorRate);
+  ovStoreFilter *filter = new ovStoreFilter(seq, maxErrorRate, beVerbose);
   ovOverlap      foverlap(seq);
   ovOverlap      roverlap(seq);
 

diff --git a/src/stores/ovStoreBuild.C b/src/stores/ovStoreBuild.C
@@ -92,6 +92,8 @@ main(int argc, char **argv) {
   bool            eValues        = false;
   char           *configOut      = NULL;
 
+  bool            beVerbose      = false;
+
   argc = AS_configure(argc, argv);
 
   vector<char *>  err;
@@ -109,6 +111,9 @@ main(int argc, char **argv) {
     } else if (strcmp(argv[arg], "-e") == 0) {
       maxErrorRate = atof(argv[++arg]);
 
+    } else if (strcmp(argv[arg], "-v") == 0) {
+      beVerbose = true;
+
     } else {
       char *s = new char [1024];
       snprintf(s, 1024, "%s: unknown option '%s'.\n", argv[0], argv[arg]);
@@ -132,6 +137,8 @@ main(int argc, char **argv) {
     fprintf(stderr, "\n");
     fprintf(stderr, "  -e e                  filter overlaps above e fraction error\n");
     fprintf(stderr, "\n");
+    fprintf(stderr, "  -v                    be overly verbose\n");
+    fprintf(stderr, "\n");
 
     for (uint32 ii=0; ii<err.size(); ii++)
       if (err[ii])
@@ -144,7 +151,7 @@ main(int argc, char **argv) {
 
   ovStoreConfig    *config = new ovStoreConfig(cfgName);
   sqStore          *seq    = sqStore::sqStore_open(seqName);
-  ovStoreFilter    *filter = new ovStoreFilter(seq, maxErrorRate);
+  ovStoreFilter    *filter = new ovStoreFilter(seq, maxErrorRate, beVerbose);
 
   //  Figure out how many overlaps there are, quit if too many.
 

diff --git a/src/stores/ovStoreFilter.C b/src/stores/ovStoreFilter.C
@@ -60,20 +60,22 @@
 
 
 
-ovStoreFilter::ovStoreFilter(sqStore *seq_, double maxErate) {
+ovStoreFilter::ovStoreFilter(sqStore *seq_, double maxErate_, bool beVerbose_) {
   seq             = seq_;
-  maxID           = seq->sqStore_getNumReads() + 1;
-  maxEvalue       = AS_OVS_encodeEvalue(maxErate);
+  maxID           = seq->sqStore_getNumReads();
+  maxEvalue       = AS_OVS_encodeEvalue(maxErate_);
+
+  beVerbose       = beVerbose_;
 
   resetCounters();
 
-  skipReadOBT     = new char [maxID];
-  skipReadDUP     = new char [maxID];
+  skipReadOBT     = new char [maxID + 1];
+  skipReadDUP     = new char [maxID + 1];
 
   uint32  numSkipOBT = 0;
   uint32  numSkipDUP = 0;
 
-  for (uint64 iid=0; iid<maxID; iid++) {
+  for (uint64 iid=0; iid<=maxID; iid++) {
     uint32     Lid = seq->sqStore_getRead(iid)->sqRead_libraryID();
     sqLibrary *L   = seq->sqStore_getLibrary(Lid);
 
@@ -154,12 +156,20 @@ void
 ovStoreFilter::filterOverlap(ovOverlap       &foverlap,
                              ovOverlap       &roverlap) {
 
+  //  GREATLY annoy the poor user that asked for 'overly verbose' mode.
+
+  if (beVerbose) {
+    char ovlstr[256];
+
+    fprintf(stderr, "%s\n", foverlap.toString(ovlstr, ovOverlapAsUnaligned, false));
+  }
+
   //  Quick sanity check on IIDs.
 
   if ((foverlap.a_iid == 0) ||
       (foverlap.b_iid == 0) ||
-      (foverlap.a_iid >= maxID) ||
-      (foverlap.b_iid >= maxID)) {
+      (foverlap.a_iid > maxID) ||
+      (foverlap.b_iid > maxID)) {
     char ovlstr[256];
 
     fprintf(stderr, "Overlap has IDs out of range (maxID " F_U32 "), possibly corrupt input data.\n", maxID);