SGSide SGLookup::mapPosition(const SGPosition& inPos) const
{
const PosMap& pm = _mapVec.at(inPos.getSeqID());
PosMap::const_iterator i = pm.lower_bound(inPos.getPos());
assert(i != pm.end());
if (i->first > inPos.getPos())
{
assert(i != pm.begin());
--i;
}
if (i->second.getBase() == SideGraph::NullPos)
{
return i->second;
}
assert(i->first <= inPos.getPos());
sg_int_t offset = inPos.getPos() - i->first;
SGSide outSide(i->second);
SGPosition outPos(outSide.getBase());
outPos.setPos(outPos.getPos() + offset);
if (outSide.getForward() == false)
{
PosMap::const_iterator j = i;
++j;
sg_int_t transform = j->first - offset - i->first - 1 - offset;
outPos.setPos(outPos.getPos() + transform);
}
outSide.setBase(outPos);
return outSide;
}
void SGLookup::addInterval(const SGPosition& inPos,
const SGPosition& outPos,
sg_int_t length,
bool reversed)
{
PosMap& pm = _mapVec.at(inPos.getSeqID());
sg_int_t left = inPos.getPos();
sg_int_t right = inPos.getPos() + length;
// find the left point
SGSide leftSide(SideGraph::NullPos, true);
PosMap::iterator li = pm.insert(pair<sg_int_t, SGSide>(
left, leftSide)).first;
// find the right point
PosMap::iterator ri = li;
++ri;
if (ri == pm.end() || ri->second.getBase().getPos() != right)
{
SGSide rightSide(SideGraph::NullPos, true);
ri = pm.insert(li, pair<sg_int_t, SGSide>(right, rightSide));
--ri;
assert(ri == li);
}
// update the left point
assert(li->second.getBase() == SideGraph::NullPos);
li->second = SGSide(outPos, !reversed);
}
void Side2Seq::convertSequence(const SGSequence* seq)
{
// we exclude the very first and last sides as they don't induce breaks
// (very first start woult have forward == true, for example..)
SGSide start = SGSide(SGPosition(seq->getID(), 0), false);
SGSide end = SGSide(SGPosition(seq->getID(), seq->getLength() - 1), true);
set<SGSide> cutSides;
if (seq->getLength() > 1)
{
getIncidentJoins(start, end, cutSides);
}
if (_chop > 0)
{
getChopSides(seq, cutSides);
}
cleanCutSides(cutSides);
SGPosition first(seq->getID(), 0);
int firstIdx = _outGraph->getNumSequences();
for (set<SGSide>::iterator i = cutSides.begin(); i != cutSides.end(); ++i)
{
SGPosition last = i->getBase();
assert(last.getSeqID() == first.getSeqID());
if (i->getForward() == true)
{
// left side of base: don't include this position
last.setPos(last.getPos() - 1);
}
// add it
addOutSequence(seq, first, last);
// add one because segments inclusive
first.setPos(last.getPos() + 1);
}
// need to do one segment at end
SGPosition last(seq->getID(), seq->getLength() - 1);
addOutSequence(seq, first, last);
// chain all the added seqeunces with new joins
for (int j = firstIdx + 1; j < _outGraph->getNumSequences(); ++j)
{
const SGSequence* fs = _outGraph->getSequence(j-1);
const SGSequence* ts = _outGraph->getSequence(j);
SGSide side1(SGPosition(fs->getID(), fs->getLength() - 1), false);
SGSide side2(SGPosition(ts->getID(), 0), true);
const SGJoin* newJoin = _outGraph->addJoin(new SGJoin(side1, side2));
verifyOutJoin(newJoin);
}
}
void Side2Seq::addOutSequence(const SGSequence* inSeq,
const SGPosition& first, const SGPosition& last)
{
int length = last.getPos() - first.getPos() + 1;
assert(length > 0);
const SGSequence* outSeq = _outGraph->addSequence(
new SGSequence(inSeq->getID(), length,
getOutSeqName(inSeq, first, length)));
_luTo.addInterval(first, SGPosition(outSeq->getID(), 0), length, false);
// add the bases
assert(_outBases.size() == outSeq->getID());
_outBases.resize(outSeq->getID() + 1);
getInDNA(SGSegment(SGSide(first, true), length), _outBases.back());
}
string Side2Seq::getOutSeqName(const SGSequence* inSeq,
const SGPosition& first,
int length) const
{
if (_stripSeqNames == true)
{
return "";
}
stringstream ss;
ss << inSeq->getName() << "_" << first.getPos();
return ss.str();
}
void SGLookup::getPath(const SGPosition& startPos,
const SGPosition& endPos,
vector<SGSegment>& outPath) const
{
SGPosition halStart = startPos;
SGPosition halEnd = endPos;
bool backward = endPos < startPos;
if (backward == true)
{
// always make a forward path. if query is reversed, we will
// remember here and flip at very end.
swap(halStart, halEnd);
}
assert(halStart.getSeqID() == halEnd.getSeqID());
const PosMap& pm = _mapVec.at(halStart.getSeqID());
// find marker to the right of halStart
PosMap::const_iterator i = pm.lower_bound(halStart.getPos());
assert(i != pm.end());
if (i->first == halStart.getPos())
{
++i;
}
// find marker >= halEnd (ie one-past what we iterate)
PosMap::const_iterator j = pm.lower_bound(halEnd.getPos());
assert(j != pm.begin());
if (j->first == halEnd.getPos())
{
++j;
}
outPath.clear();
sg_int_t pathLength = 0;
sg_int_t prevHalPos = halStart.getPos();
SGSide prevSgSide = mapPosition(halStart);
for (PosMap::const_iterator k = i; k != j; ++k)
{
assert(k != pm.end());
assert(k != pm.begin());
sg_int_t halPos = k->first;
sg_int_t segLen = halPos - prevHalPos;
assert(segLen > 0);
SGSide sgSide = k->second;
// note we are taking advantage of the fact that
// the side returned by mapPosition has a forward flag
// consistent with its use in sgsegment.
if (prevSgSide.getForward() == false && k != i)
{
// interval lookup always stores intervals left->right.
// for reverse mapping, we have to manually adjust segment
// coordinate (unless first iteration, which is already adjusted
// by call to mapPosition)
prevSgSide.setBase(SGPosition(prevSgSide.getBase().getSeqID(),
prevSgSide.getBase().getPos() + segLen -1));
}
outPath.push_back(SGSegment(prevSgSide, segLen));
pathLength += segLen;
prevHalPos = halPos;
prevSgSide = sgSide;
}
sg_int_t segLen = halEnd.getPos() - prevHalPos + 1;
assert(segLen > 0);
if (prevSgSide.getForward() == false)
{
prevSgSide.setBase(SGPosition(prevSgSide.getBase().getSeqID(),
prevSgSide.getBase().getPos() + segLen -1));
}
outPath.push_back(SGSegment(prevSgSide, segLen));
pathLength += segLen;
(void)pathLength;
if (halStart < halEnd)
{
assert(pathLength == halEnd.getPos() - halStart.getPos() + 1);
}
// we really wanted our path in the other direction. flip the
// order of the vector, and the orientation of every segment.
if (backward == true)
{
reverse(outPath.begin(), outPath.end());
for (size_t i = 0; i < outPath.size(); ++i)
{
outPath[i].flip();
}
}
}