feat: barcode support to the string graph

chungongyu · Nov 9, 2020 · 766dec1 · 766dec1
1 parent 418d727
commit 766dec1
Show file tree

Hide file tree

Showing 5 changed files with 46 additions and 7 deletions.
diff --git a/src/assembler.cpp b/src/assembler.cpp
@@ -92,7 +92,7 @@ class Assembler : public Runner {
  // Trimming
  size_t trimRound = 0, numTrimRounds = options.get<size_t>("cut-terminal", 10);
  while (trimRound < numTrimRounds) {
- LOG4CXX_INFO(logger, boost::format("[Trim] Trim round: %d") % (trimRound + 1))
+ LOG4CXX_INFO(logger, boost::format("[Trim] Trim round: %d") % (trimRound + 1));
  bool modified = false;
 
  LOG4CXX_INFO(logger, "Trimming tips");
@@ -118,7 +118,7 @@ class Assembler : public Runner {
  // Trimming
  size_t trimRound = 0, numTrimRounds = options.get<size_t>("cut-terminal", 10);
  while (trimRound < numTrimRounds) {
- LOG4CXX_INFO(logger, boost::format("[Trim] Trim round: %d") % (trimRound + 1))
+ LOG4CXX_INFO(logger, boost::format("[Trim] Trim round: %d") % (trimRound + 1));
  bool modified = false;
 
  LOG4CXX_INFO(logger, "Removing loops");

diff --git a/src/bigraph.cpp b/src/bigraph.cpp
@@ -66,8 +66,8 @@ void Edge::validate() const {
  LOG4CXX_ERROR(logger, "Error, matching strings are not the same length");
  LOG4CXX_ERROR(logger, boost::format("V1M: %s,%s") % start()->id() % m1);
  LOG4CXX_ERROR(logger, boost::format("V2M: %s,%s") % end()->id() % m2);
- LOG4CXX_ERROR(logger, boost::format("V1: %s") % v1)
- LOG4CXX_ERROR(logger, boost::format("V2: %s") % v2)
+ LOG4CXX_ERROR(logger, boost::format("V1: %s") % v1);
+ LOG4CXX_ERROR(logger, boost::format("V2: %s") % v2);
  assert(false);
  }
 }
@@ -503,7 +503,7 @@ bool Bigraph::load(std::istream& stream, size_t minOverlap, bool allowContainmen
  LOG4CXX_ERROR(logger, boost::format("Error: Unexpected vertex record found at line %s") % line);
  return false;
  }
- Vertex* vertex = new Vertex(record.id, record.seq, record.substring ? (int)record.substring : false);
+ Vertex* vertex = new Vertex(record.id, record.seq, record.substring ? (int)record.substring : false, record.barcode ? (std::string)record.barcode : "");
  if (!g->addVertex(vertex)) {
  LOG4CXX_ERROR(logger, boost::format("Error: Attempted to insert vertex into graph with a duplicate id: %s") % vertex->id());
  LOG4CXX_ERROR(logger, "All reads must have a unique identifier");

diff --git a/src/bigraph.h b/src/bigraph.h
@@ -111,7 +111,7 @@ class Vertex {
 public:
  typedef std::string Id;
 
- Vertex(const Id& id, const std::string& seq, bool contained = false) : _id(id), _seq(seq), _contained(contained), _color(GC_NONE), _coverage(1) {
+ Vertex(const Id& id, const std::string& seq, bool contained=false, const std::string& index="") : _id(id), _seq(seq), _contained(contained), _index(index), _color(GC_NONE), _coverage(1) {
  }
  ~Vertex();
 
@@ -121,6 +121,9 @@ class Vertex {
  const std::string& seq() const {
  return _seq;
  }
+ const std::string& index() const {
+ return _index;
+ }
  size_t coverage() const {
  return _coverage;
  }
@@ -173,6 +176,7 @@ class Vertex {
  Id _id;
  GraphColor _color;
  std::string _seq;
+ std::string _index;
  size_t _coverage; // Number of vertices that have been merged into this one
  bool _contained;
 

diff --git a/src/subgraph.cpp b/src/subgraph.cpp
@@ -50,7 +50,7 @@ class SubgraphExtractor {
  void addVertex(const Vertex* vertex, Bigraph* sub) {
  // Make sure the vertex hasn't been added yet
  if (sub->getVertex(vertex->id()) == NULL) {
- sub->addVertex(new Vertex(vertex->id(), vertex->seq(), vertex->contained()));
+ sub->addVertex(new Vertex(vertex->id(), vertex->seq(), vertex->contained(), vertex->index()));
  }
  }
 

diff --git a/test/index_test.cpp b/test/index_test.cpp
@@ -45,4 +45,39 @@ BOOST_AUTO_TEST_CASE(RLUnit_test) {
  }
 }
 
+BOOST_AUTO_TEST_CASE(RLString_test) {
+ DNASeq read("test", "AAACGGGTA");
+
+ RLString runs;
+
+ RLUnit run;
+ for (const auto& c : read.seq) {
+ if (run.initialized()) {
+ if (run == c && !run.full()) {
+ ++run;
+ } else {
+ runs.push_back(run);
+ run = RLUnit(c);
+ }
+ } else {
+ run = RLUnit(c);
+ }
+ }
+ if (run.initialized()) {
+ runs.push_back(run);
+ }
+
+ BOOST_CHECK_EQUAL(runs.size(), 5);
+ BOOST_CHECK_EQUAL(runs[0], 'A');
+ BOOST_CHECK_EQUAL(runs[0].count(), 3);
+ BOOST_CHECK_EQUAL(runs[1], 'C');
+ BOOST_CHECK_EQUAL(runs[1].count(), 1);
+ BOOST_CHECK_EQUAL(runs[2], 'G');
+ BOOST_CHECK_EQUAL(runs[2].count(), 3);
+ BOOST_CHECK_EQUAL(runs[3], 'T');
+ BOOST_CHECK_EQUAL(runs[3].count(), 1);
+ BOOST_CHECK_EQUAL(runs[4], 'A');
+ BOOST_CHECK_EQUAL(runs[4].count(), 1);
+}
+
 BOOST_AUTO_TEST_SUITE_END();