Add functions to get non-contained first/last read and standardize th…

…eir usage
TankMermaid · Nov 13, 2020 · ab08d0c · ab08d0c
1 parent a58af29
commit ab08d0c
Show file tree

Hide file tree

Showing 4 changed files with 40 additions and 37 deletions.
diff --git a/src/bogart/AS_BAT_FindCircular.C b/src/bogart/AS_BAT_FindCircular.C
@@ -233,8 +233,8 @@ findCircularContigs(TigVector &tigs,
     //  Grab the first and last reads in the tig, then find the edge that
     //  points out of the tig.
 
-    ufNode          *fRead = tig->firstRead();
-    ufNode          *lRead = tig->lastRead();
+    ufNode          *fRead = tig->firstBackboneRead();
+    ufNode          *lRead = tig->lastBackboneRead();
 
     uint32      circularLength = 0;
     uint32      ovlLen = 0;

diff --git a/src/bogart/AS_BAT_MergeOrphans.C b/src/bogart/AS_BAT_MergeOrphans.C
@@ -144,35 +144,6 @@ bool isCycle(TigVector       &tigs,
    return false;
 }
 
-ufNode* findFirstRead(Unitig *tig) {
-   ufNode   *read = tig->firstRead();
-
-   for (uint32 fi=1; fi < tig->ufpath.size(); fi++) {
-      if (OG->isBackbone(read->ident) && read->position.min() == 0)
-         break;
-      if (tig->ufpath[fi].position.min() == 0)
-         read=&tig->ufpath[fi];
-   }
-   assert(read->position.min() == 0);
-
-   return read;
-}
-
-ufNode* findLastRead(Unitig *tig) {
-   ufNode  *read = tig->lastRead();
-
-   for (uint32 fi=tig->ufpath.size()-1; (fi-- > 0); ) {
-      if (OG->isBackbone(read->ident) && read->position.max() == tig->getLength())
-         break;
-      if (tig->ufpath[fi].position.max() == tig->getLength())
-         read=&tig->ufpath[fi];
-   }
-   assert(read->position.max() == tig->getLength());
-
-   return read;
-}
-
-
 //  Decide which tigs can be orphans.  Any unitig where (nearly) every dovetail
 //  read has an overlap to some other unitig is a candidate for orphan popping.
 //
@@ -204,8 +175,8 @@ findPotentialOrphans(TigVector       &tigs,
 
     //  If the first or last read has no best edge, that's it, we're done.
 
-    ufNode   *fRead = findFirstRead(tig);
-    ufNode   *lRead = findLastRead(tig);
+    ufNode   *fRead = tig->firstBackboneRead();
+    ufNode   *lRead = tig->lastBackboneRead();
 
     //  Count the number of reads that have an overlap to some other tig.  tigOlapsTo[otherTig] = count.
 
@@ -941,7 +912,7 @@ mergeOrphans(TigVector    &tigs,
 
     //  Scan the orphan, decide if there are _ANY_ read placements.  Log appropriately.
 
-    if (placeAnchor(orphan, placed, findFirstRead(orphan), findLastRead(orphan)) == false) {
+    if (placeAnchor(orphan, placed, orphan->firstBackboneRead(), orphan->lastBackboneRead()) == false) {
       writeLog("\n");
       writeLog("ANCHOR READS FAILED TO PLACE.\n");
       continue;
@@ -953,8 +924,8 @@ mergeOrphans(TigVector    &tigs,
     //    read        -------
     //    orphan      -------------------------
 
-    ufNode  *fRead = findFirstRead(orphan);
-    ufNode  *lRead = findLastRead(orphan);
+    ufNode  *fRead = orphan->firstBackboneRead();
+    ufNode  *lRead = orphan->lastBackboneRead();
 
     map<uint32, intervalList<int32> *>   targetIntervals;
 
@@ -1105,6 +1076,7 @@ mergeOrphans(TigVector    &tigs,
       for (uint32 fi=0; fi<orphan->ufpath.size(); fi++) {                         //  Flag them as being an orphan, and reset backbone
         OG->setOrphan(orphan->ufpath[fi].ident);                                  //  status - they're not part of the backbone
         OG->setBackbone(orphan->ufpath[fi].ident, false);                         //  in the tig they've been placed into.
+        orphan->ufpath[fi].contained=true;
       }
 
       tigs[orphan->id()] = NULL;                                                  //  Delete the original tig.

diff --git a/src/bogart/AS_BAT_PlaceContains.C b/src/bogart/AS_BAT_PlaceContains.C
@@ -222,7 +222,7 @@ placeUnplacedUsingAllOverlaps(TigVector           &tigs,
 
     if (tig) {
       frg.ident             = fid;
-      frg.contained         = 0;
+      frg.contained         = OG->isContained(fid);
       frg.parent            = 0;
       frg.ahang             = 0;
       frg.bhang             = 0;

diff --git a/src/bogart/AS_BAT_Unitig.H b/src/bogart/AS_BAT_Unitig.H
@@ -30,6 +30,7 @@
 
 using namespace std;
 
+#define MAX_BASES_UNCONTAINED 150
 
 class BestEdgeOverlap;
 class optPos;
@@ -354,6 +355,36 @@ public:
     return(rd3);
   };
 
+  // return first or last read that is not contained or merged closest to the end of the tig
+  // allowing wiggle within start/end because of optimize position which may make a conatined read the only one touching
+  ufNode* firstBackboneRead() {
+    ufNode   *read = firstRead();
+
+    for (uint32 fi=1; fi < ufpath.size(); fi++) {
+      if (read->contained == false && read->position.min() == 0)
+        break;
+      if (ufpath[fi].position.min() <= MAX_BASES_UNCONTAINED)
+        read=&ufpath[fi];
+    }
+    assert(read->position.min() <= MAX_BASES_UNCONTAINED);
+
+    return read;
+  };
+
+  ufNode* lastBackboneRead() {
+    ufNode  *read = lastRead();
+
+    for (uint32 fi=ufpath.size()-1; (fi-- > 0); ) {
+      if (read->contained == false && read->position.max() == getLength())
+        break;
+      if (ufpath[fi].position.max() + MAX_BASES_UNCONTAINED >= getLength())
+        read=&ufpath[fi];
+    }
+    assert(read->position.max() + MAX_BASES_UNCONTAINED >= getLength());
+
+    return read;
+  };
+
   // Public Member Variables
 public:
   vector<ufNode>     ufpath;