// Construct the extension string corresponding to the path
std::string SGWalk::getString(SGWalkType type) const
{
std::string out;
if(type == SGWT_START_TO_END || type == SGWT_INTERNAL)
{
out.append(m_pStartVertex->getSeq().toString());
}
// Determine if the string should go to the end of the last vertex
// in the path
size_t stop = m_edges.size();
// The first edge is always in correct frame of reference
// so the comp is EC_SAME. This variable tracks where the
// string that is being added is different from the starting sequence
// and needs to be flipped
EdgeComp currComp = EC_SAME;
// If the walk direction is antisense, we reverse every component and then
// reverse the entire string to generate the final string
bool reverseAll = !m_edges.empty() && m_edges[0]->getDir() == ED_ANTISENSE;
if(reverseAll)
out = reverse(out);
for(size_t i = 0; i < stop; ++i)
{
Edge* pYZ = m_edges[i];
// Append in the extension string
std::string edge_str = pYZ->getLabel();
assert(edge_str.size() != 0);
if(currComp == EC_REVERSE)
edge_str = reverseComplement(edge_str);
if(reverseAll)
edge_str = reverse(edge_str);
// Calculate the next comp, between X and Z
EdgeComp ecYZ = pYZ->getComp();
EdgeComp ecXZ;
if(ecYZ == EC_SAME)
ecXZ = currComp;
else
ecXZ = !currComp;
out.append(edge_str);
currComp = ecXZ;
}
// If we want the internal portion of the string (which does not contain the endpoints
// perform the truncation now. This needs to be done before the reversal.
if(type == SGWT_INTERNAL)
{
Edge* pFirstEdge = getFirstEdge();
Edge* pLastEdge = getLastEdge();
if(pFirstEdge == NULL || pLastEdge == NULL)
{
out.clear();
}
else
{
Vertex* pStart = m_pStartVertex;
Vertex* pLast = getLastVertex();
int start = pStart->getSeqLen() - pFirstEdge->getMatchLength();
int end = out.size() - (pLast->getSeqLen() - pLastEdge->getMatchLength());
if(end <= start)
out.clear();
else
{
std::string ss = out.substr(start, end - start);
out = ss;
}
}
}
if(out.empty())
std::cout << "No output for walk: " << pathSignature() << "\n";
if(reverseAll)
out = reverse(out);
// truncate
return out;
}
// Construct the extension string corresponding to the path
std::string SGWalk::getString(SGWalkType type, SGWalkVertexPlacementVector* pPlacementVector) const
{
std::string out;
// Append the full length of the starting vertex to the walk
if(type == SGWT_START_TO_END || type == SGWT_INTERNAL)
{
out.append(m_pStartVertex->getSeq().toString());
// Add the first record to the placement vector if required
if(pPlacementVector != NULL)
{
SGWalkVertexPlacement firstPlace;
firstPlace.pVertex = m_pStartVertex;
firstPlace.position = 0; // 0-based coordinates
firstPlace.isRC = false;
pPlacementVector->push_back(firstPlace);
}
}
// Determine if the string should go to the end of the last vertex
// in the path
size_t stop = m_edges.size();
// The first edge is always in correct frame of reference
// so the comp is EC_SAME. This variable tracks where the
// string that is being added is different from the starting sequence
// and needs to be flipped
EdgeComp currComp = EC_SAME;
// If the walk direction is antisense, we reverse every component and then
// reverse the entire string to generate the final string
bool reverseAll = !m_edges.empty() && m_edges[0]->getDir() == ED_ANTISENSE;
if(reverseAll)
out = reverse(out);
for(size_t i = 0; i < stop; ++i)
{
Edge* pYZ = m_edges[i];
// Append in the extension string
std::string edge_str = pYZ->getLabel();
assert(edge_str.size() != 0);
// Determine whether this node is reverse complement wrt the string
// we are building
if(currComp == EC_REVERSE)
edge_str = reverseComplement(edge_str);
if(reverseAll)
edge_str = reverse(edge_str);
// Calculate the next comp, between X and Z
EdgeComp ecYZ = pYZ->getComp();
EdgeComp ecXZ;
if(ecYZ == EC_SAME)
ecXZ = currComp;
else
ecXZ = !currComp;
out.append(edge_str);
// Add this record into the placement vector
if(pPlacementVector != NULL)
{
SGWalkVertexPlacement placement;
placement.pVertex = pYZ->getEnd();
placement.isRC = ecXZ == EC_REVERSE;
placement.position = out.size() - pYZ->getEnd()->getSeqLen();
pPlacementVector->push_back(placement);
}
currComp = ecXZ;
}
// If we want the internal portion of the string (which does not contain the endpoints
// perform the truncation now. This needs to be done before the reversal.
if(type == SGWT_INTERNAL)
{
if(pPlacementVector != NULL)
{
std::cerr << "Error: Vertex placement not supported for SGWT_INTERNAL walk types\n";
exit(EXIT_FAILURE);
}
Edge* pFirstEdge = getFirstEdge();
Edge* pLastEdge = getLastEdge();
if(pFirstEdge == NULL || pLastEdge == NULL)
{
out.clear();
}
else
{
Vertex* pStart = m_pStartVertex;
Vertex* pLast = getLastVertex();
int start = pStart->getSeqLen() - pFirstEdge->getMatchLength();
int end = out.size() - (pLast->getSeqLen() - pLastEdge->getMatchLength());
if(end <= start)
out.clear();
else
{
std::string ss = out.substr(start, end - start);
out = ss;
}
}
}
if(out.empty())
std::cout << "No output for walk: " << pathSignature() << "\n";
if(reverseAll)
{
out = reverse(out);
// Reverse the placement vector too, including flipping the alignment coordinates
if(pPlacementVector != NULL)
{
std::reverse(pPlacementVector->begin(), pPlacementVector->end());
for(size_t i = 0; i < pPlacementVector->size(); ++i)
{
SGWalkVertexPlacement& item = pPlacementVector->at(i);
item.position = out.size() - item.position - item.pVertex->getSeqLen();
}
}
}
return out;
}
