fix traversal

inc/Kcontainer.h

@@ -46,6 +46,21 @@ inline void free_kcontainer( Kcontainer* kd )
     free(kd);
 }
 
+inline char* kcontainer_get_child_suffix(Vertex* v, int idx) {
+    uint256_t verts = v->pref_pres;
+    uint8_t j = 0, i = 0;
+    for(i = 0; i < 256; i++) {
+        if(verts & 0x1) {
+            //i++;
+            j++;
+        }
+        if(j > idx)
+            break;
+        verts >>= 1;
+    }
+    return deserialize_kmer(4, 1, &i);
+}
+
 inline bool kcontainer_contains( Kcontainer* kd, const char* kmer )
 {
     uint8_t* bseq = ( uint8_t* ) calloc( kd->k, sizeof( uint8_t ) );
@@ -66,7 +81,7 @@ inline void kcontainer_add( Kcontainer* kd, const char* kmer, int count )
 {
     uint8_t* bseq = ( uint8_t* ) calloc( kd->k, sizeof( uint8_t ) );
     serialize_kmer( kmer, kd->k, bseq );
-    //char* dseq = deserialize_kmer(kd->k, calc_bk(kd->k), bseq);
+    //std::cout << deserialize_kmer(kd->k, calc_bk(kd->k), bseq) << std::endl;
 #if KDICT
     vertex_insert( &( kd->v ), bseq, kd->k, 0, obj );
 #elif KSET

inc/Kcounter.h

@@ -39,6 +39,10 @@ class Kcounter
             UC* uc = &v->uc;
             int bk = calc_bk( k );
             int suffix_idx = bk * idx;
+            //std::cout << bk << "\t" << suffix_idx << "\t" << idx << std::endl;
+            //for(int i = 0; i < bk; i++) {
+            //    std::cout << (unsigned) uc->suffixes[0] << std::endl;
+            //}
             return deserialize_kmer( k, bk, &uc->suffixes[ suffix_idx ] );
         }
 
@@ -55,14 +59,6 @@ class Kcounter
         }
         char* get_child_suffix( Vertex* v, int idx )
         {
-            uint256_t verts = v->pref_pres;
-            uint8_t i = 0;
-            while(i < idx) {
-                if(verts & 0x1) {
-                    idx++;
-                }
-                verts >>= 1;
-            }
-            return deserialize_kmer(4, 1, &i);
+            return kcontainer_get_child_suffix(v, idx);
         }
 };

inc/Kdict.h

@@ -41,15 +41,7 @@ class Kdict
         }
         char* get_child_suffix( Vertex* v, int idx )
         {
-            uint256_t verts = v->pref_pres;
-            uint8_t i = 0;
-            while(i < idx) {
-                if(verts & 0x1) {
-                    idx++;
-                }
-                verts >>= 1;
-            }
-            return deserialize_kmer(4, 1, &i);
+            return kcontainer_get_child_suffix(v, idx);
         }
         Vertex* get_root() { return &kc->v; }
 };

inc/Kset.h

@@ -43,15 +43,7 @@ class Kset
         }
         char* get_child_suffix( Vertex* v, int idx )
         {
-            uint256_t verts = v->pref_pres;
-            uint8_t i = 0;
-            while(i < idx) {
-                if(verts & 0x1) {
-                    idx++;
-                }
-                verts >>= 1;
-            }
-            return deserialize_kmer(4, 1, &i);
+            return kcontainer_get_child_suffix(v, idx);
         }
         void add_seq(const char* seq, uint32_t length);
 

inc/helper.h

@@ -111,12 +111,13 @@ static char* deserialize_kmer( int k, int bk, uint8_t* bseq )
         else
         {
             bases_in_byte = bases_to_process;
-            tbkmer >>= 8 - (bases_in_byte * 2);
+            //tbkmer >>= 8 - (bases_in_byte * 2);
         }
 
+        tbkmer = bseq[ i ];
         for( j = 0; j < bases_in_byte; j++ )
         {
-            pos = ( i * 4 ) + (bases_in_byte - j - 1);
+            pos = ( i * 4 ) + j;
             kmer[ pos ] = COMP_TO_ASCII[ tbkmer & 0x3 ];
             tbkmer >>= 2;
         }

kcollections/__init__.py

@@ -15,11 +15,10 @@ def _get_kmers( self, v, k, prefix = '' ):
         for i in range( self.get_uc_size( v ) ):
             yield prefix + self.get_uc_kmer( v, k, i )
 
-        for i in range( self.get_cc_size( v ) ):
-            for j in range( self.get_cc_child_size( v, i ) ):
-                child_prefix = prefix + self.get_cc_child_suffix( v, i, j )
-                for kmer in self._get_kmers( self.get_cc_child_vertex( v, i, j ), k - 4, child_prefix ):
-                    yield kmer
+        for i in range( self.get_vs_size( v ) ):
+            child_prefix = prefix + self.get_child_suffix( v, i )
+            for kmer in self._get_kmers( self.get_child_vertex( v, i ), k - 4, child_prefix ):
+                yield kmer
 
     def __str__( self ):
         res = []
@@ -105,11 +104,10 @@ def _get_kmers( self, v, k, prefix = '' ):
         for i in range( self.get_uc_size( v ) ):
             yield prefix + self.get_uc_kmer( v, k, i )
 
-        for i in range( self.get_cc_size( v ) ):
-            for j in range( self.get_cc_child_size( v, i ) ):
-                child_prefix = prefix + self.get_cc_child_suffix( v, i, j )
-                for kmer in self._get_kmers( self.get_cc_child_vertex( v, i, j ), k - 4, child_prefix ):
-                    yield kmer
+        for i in range( self.get_vs_size( v ) ):
+            child_prefix = prefix + self.get_child_suffix( v, i )
+            for kmer in self._get_kmers( self.get_child_vertex( v, i ), k - 4, child_prefix ):
+                yield kmer
 
     def __str__( self ):
         return '{' + ','.join( self ) + '}'
@@ -239,11 +237,10 @@ def _get_kmers( self, v, k, prefix = '' ):
         for i in range( self.get_uc_size( v ) ):
             yield prefix + self.get_uc_kmer( v, k, i )
 
-        for i in range( self.get_cc_size( v ) ):
-            for j in range( self.get_cc_child_size( v, i ) ):
-                child_prefix = prefix + self.get_cc_child_suffix( v, i, j )
-                for kmer in self._get_kmers( self.get_cc_child_vertex( v, i, j ), k - 4, child_prefix ):
-                    yield kmer
+        for i in range( self.get_vs_size( v ) ):
+            child_prefix = prefix + self.get_child_suffix( v, i )
+            for kmer in self._get_kmers( self.get_child_vertex( v, i ), k - 4, child_prefix ):
+                yield kmer
 
     def __str__( self ):
         res = []

kcollections/src/Vertex.cc

@@ -151,6 +151,7 @@ void burst_uc( Vertex* v, int k, int depth )
         uint8_t* suffix = &bseq[ 1 ];
         uint8_t bits_to_shift = (unsigned) prefix;
         int vidx = calc_vidx(v->pref_pres, prefix);
+        //std::cout << "burst: " << deserialize_kmer(k, calc_bk(k), bseq) << "\t" << (unsigned) bseq[0] << std::endl;
 
         // check if there is already a vertex that represents this prefix
         if(!((v->pref_pres >> (unsigned) bits_to_shift) & 0x1)) {
@@ -245,8 +246,9 @@ void vertex_insert( Vertex* v, uint8_t* bseq, int k, int depth, int count )
     uc_insert( &( v->uc ), bseq, k, depth, uc_idx, count );
 #endif
 
-    if( v->uc.size > CAPACITY - 1 )
+    if(v->uc.size == CAPACITY)
     {
+        //std::cout << "bursting" << std::endl;
         burst_uc( v, k, depth );
     }
 }