feat: barcode support to the string graph

src/assembler.cpp

@@ -76,7 +76,7 @@ class Assembler : public Runner {
  delta = sigma;
  }
  }
- PairedReadVisitor prVisit(options.get<size_t>("max-distance", 100), average, delta, options.get<size_t>("max-search-nodes", 100), options.get<size_t>("threads", 1), options.get<size_t>("batch-size", 1000));
+ PairedReadVisitor prVisit(options.get<size_t>("max-distance", 100), average, delta, options.get<size_t>("max-search-nodes", 100), options.find("with-index") != options.not_found(), options.get<size_t>("threads", 1), options.get<size_t>("batch-size", 1000));
  g.visit(&prVisit);
  } else {
  LOG4CXX_INFO(logger, "Removing contained vertices from graph");
@@ -235,6 +235,7 @@ class Assembler : public Runner {
  "Paired reads parameters:\n"
  " --pe-mode=INT 0 - do not treat reads as paired (default)\n"
  " 1 - treat reads as paired\n"
+ " --with-index treat as 10x linked read data\n"
  " --max-distance=INT treat reads as connected whose distance is less than INT (default: 100)\n"
  " --insert-size=INT treat reads as paired with insert size INT (default: learned from paired reads)\n"
  " --insert-size-delta=INT treat reads as paired with insert size delta INT (default: learned from paired reads)\n"
@@ -254,7 +255,7 @@ class Assembler : public Runner {
 };
 
 static const std::string shortopts = "c:s:p:t:m:x:n:l:a:b:d:h";
-enum { OPT_HELP = 1, OPT_BATCH_SIZE, OPT_PEMODE, OPT_MAXDIST, OPT_INSERTSIZE, OPT_INSERTSIZE_DELTA, OPT_MAXEDGES, OPT_INIT_VERTEX_CAPACITY };
+enum { OPT_HELP = 1, OPT_BATCH_SIZE, OPT_PEMODE, OPT_WITH_IDX, OPT_MAXDIST, OPT_INSERTSIZE, OPT_INSERTSIZE_DELTA, OPT_MAXEDGES, OPT_INIT_VERTEX_CAPACITY };
 static const option longopts[] = {
  {"log4cxx", required_argument, NULL, 'c'}, 
  {"ini", required_argument, NULL, 's'}, 
@@ -265,6 +266,7 @@ static const option longopts[] = {
  {"threads", required_argument, NULL, 't'}, 
  {"batch-size", required_argument, NULL, OPT_BATCH_SIZE}, 
  {"pe-mode", required_argument, NULL, OPT_PEMODE}, 
+ {"with-index", no_argument, NULL, OPT_WITH_IDX}, 
  {"max-distance", required_argument, NULL, OPT_MAXDIST}, 
  {"insert-size", required_argument, NULL, OPT_INSERTSIZE}, 
  {"insert-size-delta", required_argument, NULL, OPT_INSERTSIZE_DELTA}, 

src/bigraph_visitors.cpp

@@ -1,4 +1,5 @@
 #include "bigraph_visitors.h"
+#include "asqg.h"
 #include "bigraph_search.h"
 #include "kseq.h"
 #include "reads.h"
@@ -159,6 +160,14 @@ void ContainRemoveVisitor::postvisit(Bigraph* graph) {
 //
 bool FastaVisitor::visit(Bigraph* graph, Vertex* vertex) {
  DNASeq seq(vertex->id(), vertex->seq());
+ const std::string& index = vertex->index();
+ if (!index.empty()) {
+ ASQG::StringTagValue bx(index);
+ if (!seq.comment.empty()) {
+ seq.comment += ' ';
+ }
+ seq.comment += bx.tostring(ASQG::BARCODE_TAG);
+ }
  _stream << seq;
  return false;
 }
@@ -215,7 +224,7 @@ bool LoopRemoveVisitor::visit(Bigraph* graph, Vertex* vertex) {
  Edge* nextEdge = vertex->edges(Edge::ED_SENSE)[0];
  Vertex* prevVert = prevEdge->end();
  Vertex* nextVert = nextEdge->end();
- if (prevVert == nextVert) {
+ if (!prevEdge->isSelf() && !nextEdge->isSelf() && prevVert == nextVert) {
  //vertex->color(GC_BLACK);
  _loops.push_back(vertex);
  modified = true;
@@ -589,7 +598,7 @@ class PairedVertexProcess {
  size_t numNodes[Edge::ED_COUNT] = {0}, MAX_STEPS = 3;
  for (const auto& node1 : adjacents) {
  size_t numIdx = node1->attr.distance >= 0 ? Edge::ED_SENSE : Edge::ED_ANTISENSE; 
- if (numNodes[numIdx] >= MAX_STEPS) continue;
+ //if (numNodes[numIdx] >= MAX_STEPS) continue;
  const Vertex* paired_v2 = _graph->getVertex(PairEnd::id(node1->vertex->id()));
  assert(paired_v2 != NULL);
  LOG4CXX_DEBUG(logger, boost::format("vertex1: %s<->%s, vertex2: %s<->%s") % vertex1->id() % paired_v1->id() % node1->vertex->id() % paired_v2->id());
@@ -602,6 +611,11 @@ class PairedVertexProcess {
  }, paired_v2, 0, std::abs(node1->attr.distance) + _visitor->_insertDelta*4, 1, &faraways);
  }
  for (const auto& node2 : faraways) {
+ if (_visitor->_withIndex) {
+ if (node1->vertex->index() != node2->vertex->index()) {
+ continue;
+ }
+ }
  linklist.push_back(node1);
  LOG4CXX_DEBUG(logger, boost::format("paired_read_all\t%s\t%s\t%d\t%s\t%s\t%d") % vertex1->id() % node1->vertex->id() % node1->attr.distance % paired_v1->id() % node2->vertex->id() % node2->attr.distance);
  ++numNodes[numIdx];
@@ -797,7 +811,7 @@ class PairedContainmentVisitor : public BigraphVisitor {
  for (auto& n : *containment) {
  // Modify inplace
  n->attr.distance -= distance;
- LOG4CXX_INFO(logger, boost::format("PairedContainmentVisitor::visit\t%s\t%s\t%d\t%d\t%d") % vertex->id() % n->vertex->id() % n->attr.distance % n->attr.dir % n->attr.comp);
+ LOG4CXX_DEBUG(logger, boost::format("PairedContainmentVisitor::visit\t%s\t%s\t%d\t%d\t%d") % vertex->id() % n->vertex->id() % n->attr.distance % n->attr.dir % n->attr.comp);
  }
  assert(_table.find(vertex->id()) == _table.end());
  _table[vertex->id()] = containment;

src/bigraph_visitors.h

@@ -109,7 +109,7 @@ class InsertSizeEstimateVisitor : public BigraphVisitor {
 // Visit each paired node via zigzag, sweep false positive edges.
 class PairedReadVisitor : public BigraphVisitor {
 public:
- PairedReadVisitor(size_t maxDistance, size_t insertSize, size_t insertDelta, size_t maxNodes, size_t threads=1, size_t batch=1000) : _maxDistance(maxDistance), _insertSize(insertSize), _insertDelta(insertDelta), _maxNodes(maxNodes), _threads(threads), _batch(batch) {
+ PairedReadVisitor(size_t maxDistance, size_t insertSize, size_t insertDelta, size_t maxNodes, bool withIndex, size_t threads=1, size_t batch=1000) : _maxDistance(maxDistance), _insertSize(insertSize), _insertDelta(insertDelta), _maxNodes(maxNodes), _withIndex(withIndex), _threads(threads), _batch(batch) {
  }
  void previsit(Bigraph* graph);
  bool visit(Bigraph* graph, Vertex* vertex);
@@ -121,6 +121,7 @@ class PairedReadVisitor : public BigraphVisitor {
  size_t _maxNodes;
  size_t _threads;
  size_t _batch;
+ bool _withIndex;
 
  std::vector<const Vertex *> _vertices;
  friend class PairedVertexProcess;