make including reference nodes optional

vgl-hub · Jun 21, 2024 · a4c3fd8 · a4c3fd8
1 parent 2821592
commit a4c3fd8
Show file tree

Hide file tree

Showing 6 changed files with 9 additions and 6 deletions.
diff --git a/include/input.h b/include/input.h
@@ -28,7 +28,7 @@ struct UserInputKreeq : UserInput {
     uint8_t depth = 3, backtrackingSpan = 5;
     uint64_t maxMem = 0;
     // bool
-    int doNotCollapseNodes = 0;
+    int noCollapse = 0, noReference = 0;
 
 };
 

diff --git a/src/kreeq.cpp b/src/kreeq.cpp
@@ -869,7 +869,7 @@ void DBG::DBGgraphToGFA() {
     uint32_t idCounter = 0, seqPos = 0, edgeCounter = 0;
     phmap::flat_hash_map<std::string, unsigned int>& headersToIds = *GFAsubgraph.getHash1();
 
-    if (!userInput.doNotCollapseNodes) {
+    if (!userInput.noCollapse) {
 
         phmap::parallel_flat_hash_map<uint64_t, std::tuple<DBGkmer32,uint32_t,bool>> residualEdges; // hash, kmer, G' node
 

diff --git a/src/main.cpp b/src/main.cpp
@@ -292,7 +292,8 @@ int main(int argc, char **argv) {
                 {"database", required_argument, 0, 'd'},
                 {"input-sequence", required_argument, 0, 'f'},
                 {"search-depth", required_argument, 0, 0},
-                {"do-not-collapse-nodes", no_argument, &userInput.doNotCollapseNodes, 1},
+                {"no-collapse", no_argument, &userInput.noCollapse, 1},
+                {"no-reference", no_argument, &userInput.noReference, 1},
                 {"out-format", required_argument, 0, 'o'},
                 {"input-positions", required_argument, 0, 'p'},
 
@@ -364,6 +365,8 @@ int main(int argc, char **argv) {
                         printf("\t-d --database DBG database.\n");
                         printf("\t-f --input-sequence sequence input file (fasta).\n");
                         printf("\t--search-depth the max depth for graph traversal (default: 3).\n");
+                        printf("\t--no-collapse do not collapse linear nodes (default: false).\n");
+                        printf("\t--no-reference do not include reference nodes (default: false).\n");
                         printf("\t-j --threads <n> numbers of threads (default: max).\n");
                         printf("\t-o --out-format generates various kinds of outputs (currently supported: .gfa1/2).\n");
                         printf("\t-p --input-positions BED coordinates of positions to extract kmers from.\n");

diff --git a/src/subgraph.cpp b/src/subgraph.cpp
@@ -171,7 +171,7 @@ bool DBG::DBGsubgraphFromSegment(InSegment *inSegment, std::array<uint16_t, 2> m
                         dbgKmer32color.color = 1;
                         segmentSubmap->insert(std::make_pair(got->first,dbgKmer32color));
                     }
-                }else{ // construct the kmer
+                }else if(!userInput.noReference){ // construct the kmer
 
                     DBGkmer32color dbgKmer32color;
                     dbgKmer32color.color = 2;

diff --git a/validateFiles/test.39.tst b/validateFiles/test.39.tst
@@ -1,4 +1,4 @@
-kreeq subgraph -d testFiles/random5.kreeq -f testFiles/random5.fasta --do-not-collapse-nodes 
+kreeq subgraph -d testFiles/random5.kreeq -f testFiles/random5.fasta --no-collapse
 embedded
 Subgraph summary statistics:
 Total kmers: 158

diff --git a/validateFiles/test.44.tst b/validateFiles/test.44.tst
@@ -1,4 +1,4 @@
-kreeq subgraph -d testFiles/test1.kreeq -f testFiles/random1.fasta --search-depth 0 --do-not-collapse-nodes 
+kreeq subgraph -d testFiles/test1.kreeq -f testFiles/random1.fasta --search-depth 0 --no-collapse --no-reference
 embedded
 Subgraph summary statistics:
 Total kmers: 979