void TopSegmentIsGapTest::createCallBack(Alignment *alignment) {
size_t numSequences = 3;
vector<Sequence::Info> seqVec(numSequences);
BottomSegmentIteratorPtr bi;
BottomSegmentStruct bs;
TopSegmentIteratorPtr ti;
TopSegmentStruct ts;
Genome *parent1 = alignment->addRootGenome("parent1");
Genome *child1 = alignment->addLeafGenome("child1", "parent1", 1);
// set up two genomes. each with three sequences. each sequence
// with 5 segments of length two. start with segment i in parent
// aligned with segment i in child.
for (size_t i = 0; i < numSequences; ++i) {
string name = "Sequence" + std::to_string(i);
seqVec[i] = Sequence::Info(name, 10, 5, 5);
}
parent1->setDimensions(seqVec);
child1->setDimensions(seqVec);
for (bi = parent1->getBottomSegmentIterator(); not bi->atEnd(); bi->toRight()) {
bs.set(bi->getBottomSegment()->getArrayIndex() * 2, 2);
bs._children.clear();
bs._children.push_back(pair<hal_size_t, bool>(bi->getBottomSegment()->getArrayIndex(), false));
bs.applyTo(bi);
}
for (ti = child1->getTopSegmentIterator(); not ti->atEnd(); ti->toRight()) {
ts.set(ti->getTopSegment()->getArrayIndex() * 2, 2, ti->getTopSegment()->getArrayIndex());
ts.applyTo(ti);
}
// insertion in middle (8th top segment)
bi = parent1->getBottomSegmentIterator(8);
ti = child1->getTopSegmentIterator(8);
assert(bi->getBottomSegment()->getChildIndex(0) == 8 && ti->getTopSegment()->getParentIndex() == 8);
bi->getBottomSegment()->setChildIndex(0, 9);
ti->getTopSegment()->setParentIndex(NULL_INDEX);
ti->toRight();
ti->getTopSegment()->setParentIndex(8);
// insertion at begining (10th top segment)
bi = parent1->getBottomSegmentIterator(10);
ti = child1->getTopSegmentIterator(10);
assert(bi->getBottomSegment()->getChildIndex(0) == 10 && ti->getTopSegment()->getParentIndex() == 10);
bi->getBottomSegment()->setChildIndex(0, 11);
ti->getTopSegment()->setParentIndex(NULL_INDEX);
ti->toRight();
ti->getTopSegment()->setParentIndex(10);
// just having a null parent is not enough for an insertion
bi = parent1->getBottomSegmentIterator(2);
ti = child1->getTopSegmentIterator(2);
assert(bi->getBottomSegment()->getChildIndex(0) == 2 && ti->getTopSegment()->getParentIndex() == 2);
ti->getTopSegment()->setParentIndex(NULL_INDEX);
}
void MappedSegmentMapDupeTest::createCallBack(AlignmentPtr alignment)
{
vector<Sequence::Info> seqVec(1);
BottomSegmentIteratorPtr bi;
BottomSegmentStruct bs;
TopSegmentIteratorPtr ti;
TopSegmentStruct ts;
// setup simple case were there is an edge from a parent to
// child and it is reversed
Genome* parent = alignment->addRootGenome("parent");
Genome* child1 = alignment->addLeafGenome("child1", "parent", 1);
Genome* child2 = alignment->addLeafGenome("child2", "parent", 1);
seqVec[0] = Sequence::Info("Sequence", 3, 0, 1);
parent->setDimensions(seqVec);
seqVec[0] = Sequence::Info("Sequence", 9, 3, 0);
child1->setDimensions(seqVec);
seqVec[0] = Sequence::Info("Sequence", 9, 3, 0);
child2->setDimensions(seqVec);
parent->setString("CCC");
child1->setString("CCCTACGTG");
child2->setString("CCCTACGTG");
bi = parent->getBottomSegmentIterator();
bs.set(0, 3);
bs._children.push_back(pair<hal_size_t, bool>(0, true));
bs._children.push_back(pair<hal_size_t, bool>(0, false));
bs.applyTo(bi);
ti = child1->getTopSegmentIterator();
ts.set(0, 3, 0, true, NULL_INDEX, 1);
ts.applyTo(ti);
ti->toRight();
ts.set(3, 3, 0, true, NULL_INDEX, 2);
ts.applyTo(ti);
ti->toRight();
ts.set(6, 3, 0, true, NULL_INDEX, 0);
ts.applyTo(ti);
ti = child2->getTopSegmentIterator();
ts.set(0, 3, 0, false);
ts.applyTo(ti);
ti->toRight();
ts.set(3, 3, NULL_INDEX, true);
ts.applyTo(ti);
ti->toRight();
ts.set(6, 3, NULL_INDEX, false);
ts.applyTo(ti);
}
void TopSegmentSimpleIteratorTest::checkCallBack(const Alignment *alignment) {
const Genome *ancGenome = alignment->openGenome("Anc0");
CuAssertTrue(_testCase, ancGenome->getNumTopSegments() == _topSegments.size());
TopSegmentIteratorPtr tsIt = ancGenome->getTopSegmentIterator(0);
for (size_t i = 0; i < ancGenome->getNumTopSegments(); ++i) {
CuAssertTrue(_testCase, (size_t)tsIt->getTopSegment()->getArrayIndex() == i);
_topSegments[i].compareTo(tsIt, _testCase);
tsIt->toRight();
}
tsIt = ancGenome->getTopSegmentIterator(ancGenome->getNumTopSegments() - 1);
for (hal_index_t i = ancGenome->getNumTopSegments() - 1; i >= 0; --i) {
CuAssertTrue(_testCase, tsIt->getTopSegment()->getArrayIndex() == i);
_topSegments[i].compareTo(tsIt, _testCase);
tsIt->toLeft();
}
tsIt = ancGenome->getTopSegmentIterator(0);
tsIt->slice(0, tsIt->getLength() - 1);
for (hal_index_t i = 0; i < (hal_index_t)ancGenome->getSequenceLength(); ++i) {
CuAssertTrue(_testCase, tsIt->getLength() == 1);
CuAssertTrue(_testCase, tsIt->getStartPosition() == i);
tsIt->toRight(tsIt->getStartPosition() + 1);
}
tsIt = ancGenome->getTopSegmentIterator(ancGenome->getNumTopSegments() - 1);
tsIt->slice(tsIt->getLength() - 1, 0);
for (hal_index_t i = ancGenome->getSequenceLength() - 1; i >= 0; --i) {
CuAssertTrue(_testCase, tsIt->getLength() == 1);
CuAssertTrue(_testCase, tsIt->getStartPosition() == i);
tsIt->toLeft(tsIt->getStartPosition() - 1);
}
tsIt = ancGenome->getTopSegmentIterator(0);
tsIt->toReverse();
CuAssertTrue(_testCase, tsIt->getReversed() == true);
tsIt->slice(tsIt->getLength() - 1, 0);
for (hal_index_t i = 0; i < (hal_index_t)ancGenome->getSequenceLength(); ++i) {
CuAssertTrue(_testCase, tsIt->getLength() == 1);
CuAssertTrue(_testCase, tsIt->getStartPosition() == i);
tsIt->toLeft(tsIt->getStartPosition() + 1);
}
tsIt = ancGenome->getTopSegmentIterator(ancGenome->getNumTopSegments() - 1);
tsIt->toReverse();
tsIt->slice(0, tsIt->getLength() - 1);
for (hal_index_t i = ancGenome->getSequenceLength() - 1; i >= 0; --i) {
CuAssertTrue(_testCase, tsIt->getLength() == 1);
CuAssertTrue(_testCase, tsIt->getStartPosition() == i);
tsIt->toRight(tsIt->getStartPosition() - 1);
}
}
void TopSegmentSimpleIteratorTest::createCallBack(Alignment *alignment) {
Genome *ancGenome = alignment->addRootGenome("Anc0", 0);
size_t numChildren = 9;
for (size_t i = 0; i < numChildren; ++i) {
alignment->addLeafGenome("Leaf" + std::to_string(i), "Anc0", 0.1);
}
vector<Sequence::Info> seqVec(1);
seqVec[0] = Sequence::Info("Sequence", 1000000, 5000, 10000);
ancGenome->setDimensions(seqVec);
CuAssertTrue(_testCase, ancGenome->getNumChildren() == numChildren);
_topSegments.clear();
for (size_t i = 0; i < ancGenome->getNumTopSegments(); ++i) {
TopSegmentStruct topSeg;
topSeg.setRandom();
topSeg._length = ancGenome->getSequenceLength() / ancGenome->getNumTopSegments();
topSeg._startPosition = i * topSeg._length;
_topSegments.push_back(topSeg);
}
TopSegmentIteratorPtr tsIt = ancGenome->getTopSegmentIterator(0);
for (size_t i = 0; not tsIt->atEnd(); tsIt->toRight(), ++i) {
CuAssertTrue(_testCase, (size_t)tsIt->getTopSegment()->getArrayIndex() == i);
_topSegments[i].applyTo(tsIt);
}
}
void GappedSegmentSimpleIteratorTest2::createCallBack(AlignmentPtr alignment)
{
addIdenticalParentChild(alignment, 2, 100, 5);
Genome* parent = alignment->openGenome(alignment->getRootName());
Genome* child = parent->getChild(0);
TopSegmentIteratorPtr ti = child->getTopSegmentIterator();
BottomSegmentIteratorPtr bi = parent->getBottomSegmentIterator();
hal_index_t i = 0;
bool reversed = true;
while (ti != child->getTopSegmentEndIterator())
{
if (i % 5 == 0)
{
reversed = !reversed;
if (reversed && i < (hal_index_t)(parent->getNumBottomSegments() - 1))
{
makeInversion(ti, 5);
}
}
ti->toRight();
bi->toRight();
++i;
}
}
void SequenceIteratorTest::checkCallBack(const Alignment *alignment) {
const Genome *ancGenome = alignment->openGenome("AncGenome");
hal_size_t numSequences = ancGenome->getNumSequences();
CuAssertTrue(_testCase, numSequences = 1000);
for (SequenceIteratorPtr seqIt = ancGenome->getSequenceIterator(); not seqIt->atEnd(); seqIt->toNext()) {
const Sequence *seq = seqIt->getSequence();
hal_size_t i = seq->getArrayIndex();
TopSegmentIteratorPtr tsIt = seq->getTopSegmentIterator();
hal_size_t numTopSegments = seq->getNumTopSegments();
for (hal_size_t j = 0; j < numTopSegments; ++j) {
TopSegmentIteratorPtr gtsIt = ancGenome->getTopSegmentIterator((i - 1) * 100 + j);
const TopSegment *gsTopSegment = gtsIt->getTopSegment();
const TopSegment *sqTopSegment = tsIt->getTopSegment();
CuAssertTrue(_testCase, gsTopSegment->getArrayIndex() == sqTopSegment->getArrayIndex());
tsIt->toRight();
}
BottomSegmentIteratorPtr bsIt = seq->getBottomSegmentIterator();
hal_size_t numBottomSegments = seq->getNumBottomSegments();
for (hal_size_t j = 0; j < numBottomSegments; ++j) {
BottomSegmentIteratorPtr gbsIt = ancGenome->getBottomSegmentIterator((i - 1) * 100 + j);
const BottomSegment *gsBottomSegment = gbsIt->getBottomSegment();
const BottomSegment *sqBottomSegment = bsIt->getBottomSegment();
CuAssertTrue(_testCase, gsBottomSegment->getArrayIndex() == sqBottomSegment->getArrayIndex());
bsIt->toRight();
}
}
}
void Genome::fixParseInfo()
{
if (getParent() == NULL || getNumChildren() == 0)
{
return;
}
// copied from CactusHalConverter::updateRootParseInfo() in
// cactus2hal/src/cactusHalConverter.cpp
BottomSegmentIteratorPtr bottomIterator =
getBottomSegmentIterator();
TopSegmentIteratorPtr topIterator = getTopSegmentIterator();
BottomSegmentIteratorConstPtr bend = getBottomSegmentEndIterator();
TopSegmentIteratorConstPtr tend = getTopSegmentEndIterator();
int top = 0, bot = 0;
while (bottomIterator != bend && topIterator != tend)
{
bool bright = false;
bool tright = false;
BottomSegment* bseg = bottomIterator->getBottomSegment();
TopSegment* tseg = topIterator->getTopSegment();
hal_index_t bstart = bseg->getStartPosition();
hal_index_t bendidx = bstart + (hal_index_t)bseg->getLength();
hal_index_t tstart = tseg->getStartPosition();
hal_index_t tendidx = tstart + (hal_index_t)tseg->getLength();
if (bstart >= tstart && bstart < tendidx)
{
bseg->setTopParseIndex(tseg->getArrayIndex());
}
if (bendidx <= tendidx || bstart == bendidx)
{
bright = true;
}
if (tstart >= bstart && tstart < bendidx)
{
tseg->setBottomParseIndex(bseg->getArrayIndex());
}
if (tendidx <= bendidx || tstart == tendidx)
{
tright = true;
}
assert(bright || tright);
if (bright == true)
{
bot += 1;
bottomIterator->toRight();
}
if (tright == true)
{
top += 1;
topIterator->toRight();
}
}
}
void GappedBottomSegmentIterator::toRightNextUngapped(TopSegmentIteratorPtr topSeqIt) const {
while (topSeqIt->tseg()->hasParent() == false && topSeqIt->getLength() <= _gapThreshold) {
if ((!topSeqIt->getReversed() && topSeqIt->getTopSegment()->isLast()) ||
(topSeqIt->getReversed() && topSeqIt->getTopSegment()->isFirst())) {
break;
}
topSeqIt->toRight();
}
}
// note: takes smart pointer as it maybe added to the results
static hal_size_t mapUp(MappedSegmentPtr mappedSeg, list<MappedSegmentPtr> &results, bool doDupes, hal_size_t minLength) {
const Genome *parent = mappedSeg->getGenome()->getParent();
assert(parent != NULL);
hal_size_t added = 0;
if (mappedSeg->isTop() == true) {
BottomSegmentIteratorPtr botSegIt = parent->getBottomSegmentIterator();
TopSegmentIteratorPtr topSegIt = mappedSeg->targetAsTop();
if (topSegIt->tseg()->hasParent() == true && topSegIt->getLength() >= minLength &&
(doDupes == true || topSegIt->tseg()->isCanonicalParalog() == true)) {
botSegIt->toParent(topSegIt);
mappedSeg->setTarget(std::dynamic_pointer_cast<SegmentIterator>(botSegIt));
results.push_back(mappedSeg);
++added;
}
} else {
hal_index_t rightCutoff = mappedSeg->getEndPosition();
BottomSegmentIteratorPtr botSegIt = mappedSeg->targetAsBottom();
hal_index_t startOffset = (hal_index_t)botSegIt->getStartOffset();
hal_index_t endOffset = (hal_index_t)botSegIt->getEndOffset();
TopSegmentIteratorPtr topSegIt = mappedSeg->getGenome()->getTopSegmentIterator();
topSegIt->toParseUp(botSegIt);
do {
TopSegmentIteratorPtr newTopSegIt = topSegIt->clone();
// we map the new target back to see how the offsets have
// changed. these changes are then applied to the source segment
// as deltas
BottomSegmentIteratorPtr backBotSegIt = botSegIt->clone();
backBotSegIt->toParseDown(newTopSegIt);
hal_index_t startBack = (hal_index_t)backBotSegIt->getStartOffset();
hal_index_t endBack = (hal_index_t)backBotSegIt->getEndOffset();
assert(startBack >= startOffset);
assert(endBack >= endOffset);
SegmentIteratorPtr newSourceSegIt = mappedSeg->sourceClone();
hal_index_t startDelta = startBack - startOffset;
hal_index_t endDelta = endBack - endOffset;
assert((hal_index_t)newSourceSegIt->getLength() > startDelta + endDelta);
newSourceSegIt->slice(newSourceSegIt->getStartOffset() + startDelta, newSourceSegIt->getEndOffset() + endDelta);
MappedSegmentPtr newMappedSeg(new MappedSegment(newSourceSegIt, newTopSegIt));
assert(newMappedSeg->isTop() == true);
assert(newMappedSeg->getSource()->getGenome() == mappedSeg->getSource()->getGenome());
added += mapUp(newMappedSeg, results, doDupes, minLength);
// stupid that we have to make this check but odn't want to
// make fundamental api change now
if (topSegIt->getEndPosition() != rightCutoff) {
topSegIt->toRight(rightCutoff);
} else {
break;
}
} while (true);
}
return added;
}
// Set top segments to be equal width and so that segment 1, 2, 3,
// etc. corresponds to parent segment 1, 2, 3, etc.
void setTopSegments(Genome *genome, hal_size_t width) {
TopSegmentIteratorPtr topIt = genome->getTopSegmentIterator();
hal_size_t n = genome->getNumTopSegments();
hal_index_t startPos = 0;
for (; topIt->getArrayIndex() < n; topIt->toRight(), startPos += width) {
topIt->setCoordinates(startPos, width);
topIt->tseg()->setParentIndex(topIt->getArrayIndex());
topIt->tseg()->setParentReversed(false);
topIt->tseg()->setBottomParseIndex(NULL_INDEX);
topIt->tseg()->setNextParalogyIndex(NULL_INDEX);
}
}
void Genome::copyTopSegments(Genome *dest) const
{
const Genome *inParent = getParent();
const Genome *outParent = dest->getParent();
TopSegmentIteratorConstPtr inTop = getTopSegmentIterator();
TopSegmentIteratorPtr outTop = dest->getTopSegmentIterator();
hal_size_t n = dest->getNumTopSegments();
assert(n == 0 || n == getNumTopSegments());
if (n == 0) {
// Nothing to do if there are no top segments.
return;
}
BottomSegmentIteratorConstPtr inParentBottomSegIt = inParent->getBottomSegmentIterator();
BottomSegmentIteratorConstPtr outParentBottomSegIt = outParent->getBottomSegmentIterator();
for (; (hal_size_t)inTop->getArrayIndex() < n; inTop->toRight(),
outTop->toRight())
{
hal_index_t genomePos = inTop->getStartPosition();
assert(genomePos != NULL_INDEX);
string inSeqName = getSequenceBySite(genomePos)->getName();
string outSeqName = dest->getSequenceBySite(genomePos)->getName();
// if (inSeqName != outSeqName) {
// stringstream ss;
// ss << "When copying top segments from " << getName() << " to " << dest->getName() << ": sequence " << inSeqName << " != " << outSeqName << " at site " << genomePos;
// throw hal_exception(ss.str());
// }
outTop->setCoordinates(inTop->getStartPosition(), inTop->getLength());
outTop->setParentIndex(inTop->getParentIndex());
outTop->setParentReversed(inTop->getParentReversed());
outTop->setBottomParseIndex(inTop->getBottomParseIndex());
outTop->setNextParalogyIndex(inTop->getNextParalogyIndex());
// Check that the sequences from the bottom segments we point to are the same. If not, correct the indices so that they are.
if (inTop->getParentIndex() != NULL_INDEX) {
inParentBottomSegIt->toParent(inTop);
const Sequence *inParentSequence = inParentBottomSegIt->getSequence();
const Sequence *outParentSequence = outParent->getSequence(inParentSequence->getName());
hal_index_t inParentSegmentOffset = inTop->getParentIndex() - inParentSequence->getBottomSegmentArrayIndex();
hal_index_t outParentSegmentIndex = inParentSegmentOffset + outParentSequence->getBottomSegmentArrayIndex();
outTop->setParentIndex(outParentSegmentIndex);
}
}
}
int MappedSegment::boundComp(const SegmentIteratorPtr &s1, const SegmentIteratorPtr &s2) {
int res = 0;
bool flip = s2->getReversed();
if (flip) {
s2->toReverse();
}
if (s1->isTop() && !s2->isTop()) {
BottomSegmentIteratorPtr bot = std::dynamic_pointer_cast<BottomSegmentIterator>(s2);
hal_index_t lb = bot->bseg()->getTopParseIndex();
hal_index_t ub = lb;
if ((hal_size_t)bot->getArrayIndex() < bot->getGenome()->getNumBottomSegments() - 1) {
bot = bot->clone();
bot->slice(0, 0);
bot->toRight();
ub = bot->bseg()->getTopParseIndex();
}
if (s1->getArrayIndex() < lb) {
res = -1;
} else if (s1->getArrayIndex() > ub) {
res = 1;
}
} else if (!s1->isTop() && s2->isTop()) {
TopSegmentIteratorPtr top = std::dynamic_pointer_cast<TopSegmentIterator>(s2);
hal_index_t lb = top->tseg()->getBottomParseIndex();
hal_index_t ub = lb;
if ((hal_size_t)top->getArrayIndex() < top->getGenome()->getNumTopSegments() - 1) {
top = top->clone();
top->slice(0, 0);
top->toRight();
ub = top->tseg()->getBottomParseIndex();
}
if (s1->getArrayIndex() < lb) {
res = -1;
} else if (s1->getArrayIndex() > ub) {
res = 1;
}
}
if (flip) {
s2->toReverse();
}
return res;
}
void GappedSegmentSimpleIteratorTest::createCallBack(AlignmentPtr alignment)
{
addIdenticalParentChild(alignment, 2, 100, 5);
Genome* parent = alignment->openGenome(alignment->getRootName());
Genome* child = parent->getChild(0);
TopSegmentIteratorPtr ti = child->getTopSegmentIterator();
BottomSegmentIteratorPtr bi = parent->getBottomSegmentIterator();
int i = 0;
while (ti != child->getTopSegmentEndIterator())
{
if (i++ % 2)
{
ti->getTopSegment()->setParentReversed(true);
bi->getBottomSegment()->setChildReversed(0, true);
}
ti->toRight();
bi->toRight();
}
}
void TopSegmentIteratorToSiteTest::createCallBack(AlignmentPtr alignment)
{
vector<Sequence::Info> seqVec(1);
TopSegmentIteratorPtr ti;
TopSegmentStruct ts;
// case 1: single segment
Genome* case1 = alignment->addRootGenome("case1");
seqVec[0] = Sequence::Info("Sequence", 10, 2, 0);
case1->setDimensions(seqVec);
ti = case1->getTopSegmentIterator();
ts.set(0, 9);
ts.applyTo(ti);
ti->toRight();
ts.set(9, 1);
ts.applyTo(ti);
case1 = NULL;
// case 2: bunch of random segments
const hal_size_t numSegs = 1133;
hal_size_t total = 0;
vector<hal_size_t> segLens(numSegs);
for (size_t i = 0 ; i < numSegs; ++i)
{
hal_size_t len = rand() % 77 + 1;
segLens[i] = len;
total += len;
assert(len > 0);
}
Genome* case2 = alignment->addRootGenome("case2");
seqVec[0] = Sequence::Info("Sequence", total, numSegs, 0);
case2->setDimensions(seqVec);
hal_index_t prev = 0;
for (size_t i = 0 ; i < numSegs; ++i)
{
ti = case2->getTopSegmentIterator((hal_index_t)i);
ts.set(prev, segLens[i]);
prev += segLens[i];
ts.applyTo(ti);
}
}
void GenomeCopyTest::checkCallBack(const Alignment *alignment) {
// FIXME: halAlignment->open() fails miserably but
// openHalAlignmentReadOnly works? Probably some state isn't cleared
// on close.
AlignmentPtr tmp(getTestAlignmentInstances(alignment->getStorageFormat(), _path, WRITE_ACCESS));
_secondAlignment = tmp;
const Genome *ancGenome = alignment->openGenome("AncGenome");
CuAssertTrue(_testCase, ancGenome->getName() == "AncGenome");
CuAssertTrue(_testCase, ancGenome->getSequenceLength() == 1000000);
CuAssertTrue(_testCase, ancGenome->getNumTopSegments() == 0);
CuAssertTrue(_testCase, ancGenome->getNumBottomSegments() == 700000);
const MetaData *ancMeta = ancGenome->getMetaData();
CuAssertTrue(_testCase, ancMeta->get("Young") == "Jeezy");
const Genome *leafGenome = alignment->openGenome("LeafGenome1");
string ancSeq = "CAT";
hal_index_t n = ancGenome->getSequenceLength();
DnaIteratorPtr dnaIt = ancGenome->getDnaIterator();
for (; dnaIt->getArrayIndex() < n; dnaIt->toRight()) {
size_t i = dnaIt->getArrayIndex() % ancSeq.size();
CuAssertTrue(_testCase, dnaIt->getBase() == ancSeq[i]);
}
TopSegmentIteratorPtr topIt = leafGenome->getTopSegmentIterator();
n = leafGenome->getNumTopSegments();
for (; topIt->getArrayIndex() < n; topIt->toRight()) {
CuAssertTrue(_testCase, topIt->getStartPosition() == topIt->getArrayIndex());
CuAssertTrue(_testCase, topIt->getLength() == 1);
CuAssertTrue(_testCase, topIt->tseg()->getParentIndex() == 3);
CuAssertTrue(_testCase, topIt->tseg()->getParentReversed() == true);
CuAssertTrue(_testCase, topIt->tseg()->getBottomParseIndex() == 5);
if (topIt->getArrayIndex() != 6) {
CuAssertTrue(_testCase, topIt->tseg()->getNextParalogyIndex() == 6);
} else {
CuAssertTrue(_testCase, topIt->tseg()->getNextParalogyIndex() == 7);
}
}
BottomSegmentIteratorPtr botIt = ancGenome->getBottomSegmentIterator();
n = ancGenome->getNumBottomSegments();
for (; botIt->getArrayIndex() < n; botIt->toRight()) {
CuAssertTrue(_testCase, botIt->getStartPosition() == botIt->getArrayIndex());
CuAssertTrue(_testCase, botIt->getLength() == 1);
CuAssertTrue(_testCase, botIt->bseg()->getChildIndex(0) == 3);
CuAssertTrue(_testCase, botIt->bseg()->getChildReversed(0) == true);
CuAssertTrue(_testCase, botIt->bseg()->getTopParseIndex() == 5);
}
const Genome *copyRootGenome = _secondAlignment->openGenome("copyRootGenome");
const Genome *copyLeafGenome = _secondAlignment->openGenome("LeafGenome1");
CuAssertTrue(_testCase, copyRootGenome->getName() == "copyRootGenome");
CuAssertTrue(_testCase, copyRootGenome->getSequenceLength() == 1000000);
CuAssertTrue(_testCase, copyRootGenome->getNumTopSegments() == 0);
CuAssertTrue(_testCase, copyRootGenome->getNumBottomSegments() == 700000);
CuAssertTrue(_testCase, copyLeafGenome->getName() == "LeafGenome1");
CuAssertTrue(_testCase, copyLeafGenome->getSequenceLength() == 1000000);
CuAssertTrue(_testCase, copyLeafGenome->getNumTopSegments() == 5000);
CuAssertTrue(_testCase, copyLeafGenome->getNumBottomSegments() == 0);
const MetaData *copyMeta = copyRootGenome->getMetaData();
CuAssertTrue(_testCase, copyMeta->get("Young") == "Jeezy");
n = copyRootGenome->getSequenceLength();
dnaIt = copyRootGenome->getDnaIterator();
for (; dnaIt->getArrayIndex() < n; dnaIt->toRight()) {
size_t i = dnaIt->getArrayIndex() % ancSeq.size();
CuAssertTrue(_testCase, dnaIt->getBase() == ancSeq[i]);
}
topIt = copyLeafGenome->getTopSegmentIterator();
n = copyLeafGenome->getNumTopSegments();
for (; topIt->getArrayIndex() < n; topIt->toRight()) {
CuAssertTrue(_testCase, topIt->getStartPosition() == topIt->getArrayIndex());
CuAssertTrue(_testCase, topIt->getLength() == 1);
CuAssertTrue(_testCase, topIt->tseg()->getParentIndex() == 3);
CuAssertTrue(_testCase, topIt->tseg()->getParentReversed() == true);
CuAssertTrue(_testCase, topIt->tseg()->getBottomParseIndex() == 5);
if (topIt->getArrayIndex() != 6) {
CuAssertTrue(_testCase, topIt->tseg()->getNextParalogyIndex() == 6);
} else {
CuAssertTrue(_testCase, topIt->tseg()->getNextParalogyIndex() == 7);
}
}
botIt = copyRootGenome->getBottomSegmentIterator();
n = copyRootGenome->getNumBottomSegments();
for (; botIt->getArrayIndex() < n; botIt->toRight()) {
CuAssertTrue(_testCase, botIt->getStartPosition() == botIt->getArrayIndex());
CuAssertTrue(_testCase, botIt->getLength() == 1);
CuAssertTrue(_testCase, botIt->bseg()->getChildIndex(0) == 3);
CuAssertTrue(_testCase, botIt->bseg()->getChildReversed(0) == true);
CuAssertTrue(_testCase, botIt->bseg()->getTopParseIndex() == 5);
}
_secondAlignment->close();
remove(_path.c_str());
}
void GenomeCopyTest::createCallBack(Alignment *alignment) {
hal_size_t alignmentSize = alignment->getNumGenomes();
CuAssertTrue(_testCase, alignmentSize == 0);
// Hacky: Need a different alignment to test copying the bottom
// segments correctly. (the names of a node's children are used
// when copying bottom segments, and two genomes can't have the same
// name in the same alignment)
_path = getTempFile();
_secondAlignment =
AlignmentPtr(getTestAlignmentInstances(alignment->getStorageFormat(), _path, WRITE_ACCESS | CREATE_ACCESS));
Genome *ancGenome = alignment->addRootGenome("AncGenome", 0);
Genome *leafGenome = alignment->addLeafGenome("LeafGenome1", "AncGenome", 0);
// This genome will test copyDimensions, copyTopSegments,
// copyBottomSegments, copySequence, copyMetadata
Genome *copyRootGenome = _secondAlignment->addRootGenome("copyRootGenome", 0);
Genome *copyLeafGenome = _secondAlignment->addLeafGenome("LeafGenome1", "copyRootGenome", 0);
MetaData *ancMeta = ancGenome->getMetaData();
ancMeta->set("Young", "Jeezy");
vector<Sequence::Info> seqVec(1);
seqVec[0] = Sequence::Info("Sequence", 1000000, 0, 700000);
ancGenome->setDimensions(seqVec);
seqVec[0] = Sequence::Info("Sequence", 1000000, 5000, 0);
leafGenome->setDimensions(seqVec);
string ancSeq = "CAT";
hal_index_t n = ancGenome->getSequenceLength();
DnaIteratorPtr dnaIt = ancGenome->getDnaIterator();
for (; dnaIt->getArrayIndex() < n; dnaIt->toRight()) {
size_t i = dnaIt->getArrayIndex() % ancSeq.size();
dnaIt->setBase(ancSeq[i]);
}
dnaIt->flush();
n = leafGenome->getSequenceLength();
dnaIt = leafGenome->getDnaIterator();
for (; dnaIt->getArrayIndex() < n; dnaIt->toRight()) {
size_t i = dnaIt->getArrayIndex() % ancSeq.size();
dnaIt->setBase(ancSeq[i]);
}
dnaIt->flush();
TopSegmentIteratorPtr topIt = leafGenome->getTopSegmentIterator();
n = leafGenome->getNumTopSegments();
for (; topIt->getArrayIndex() < n; topIt->toRight()) {
topIt->setCoordinates(topIt->getArrayIndex(), 1);
topIt->tseg()->setParentIndex(3);
topIt->tseg()->setParentReversed(true);
topIt->tseg()->setBottomParseIndex(5);
if (topIt->getArrayIndex() != 6) {
topIt->tseg()->setNextParalogyIndex(6);
} else {
topIt->tseg()->setNextParalogyIndex(7);
}
}
BottomSegmentIteratorPtr botIt = ancGenome->getBottomSegmentIterator();
n = ancGenome->getNumBottomSegments();
for (; botIt->getArrayIndex() < n; botIt->toRight()) {
botIt->setCoordinates(botIt->getArrayIndex(), 1);
botIt->bseg()->setChildIndex(0, 3);
botIt->bseg()->setChildReversed(0, true);
botIt->bseg()->setTopParseIndex(5);
}
seqVec[0] = Sequence::Info("Sequence", 3300, 0, 1100);
copyRootGenome->setDimensions(seqVec);
seqVec[0] = Sequence::Info("Sequence", 3300, 2200, 0);
copyLeafGenome->setDimensions(seqVec);
string copySeq = "TAG";
dnaIt = copyRootGenome->getDnaIterator();
n = copyRootGenome->getSequenceLength();
for (; dnaIt->getArrayIndex() < n; dnaIt->toRight()) {
size_t i = dnaIt->getArrayIndex() % copySeq.size();
dnaIt->setBase(copySeq[i]);
}
dnaIt->flush();
dnaIt = copyLeafGenome->getDnaIterator();
n = copyLeafGenome->getSequenceLength();
for (; dnaIt->getArrayIndex() < n; dnaIt->toRight()) {
size_t i = dnaIt->getArrayIndex() % copySeq.size();
dnaIt->setBase(copySeq[i]);
}
dnaIt->flush();
topIt = copyLeafGenome->getTopSegmentIterator();
n = copyLeafGenome->getNumTopSegments();
for (; topIt->getArrayIndex() < n; topIt->toRight()) {
topIt->setCoordinates(7, 8);
topIt->tseg()->setParentIndex(9);
topIt->tseg()->setParentReversed(false);
topIt->tseg()->setBottomParseIndex(11);
if (topIt->getArrayIndex() != 12) {
topIt->tseg()->setNextParalogyIndex(12);
} else {
topIt->tseg()->setNextParalogyIndex(7);
}
}
botIt = copyRootGenome->getBottomSegmentIterator();
n = copyRootGenome->getNumBottomSegments();
for (; botIt->getArrayIndex() < n; botIt->toRight()) {
botIt->setCoordinates(6, 7);
botIt->bseg()->setChildIndex(0, 8);
botIt->bseg()->setChildReversed(0, false);
botIt->bseg()->setTopParseIndex(10);
}
ancGenome->copy(copyRootGenome);
leafGenome->copy(copyLeafGenome);
_secondAlignment->close();
}
// note: takes smart pointer as it maybe added to the results
static hal_size_t mapSelf(MappedSegmentPtr mappedSeg, list<MappedSegmentPtr> &results, hal_size_t minLength) {
hal_size_t added = 0;
if (mappedSeg->isTop() == true) {
SegmentIteratorPtr target = mappedSeg->getTargetIteratorPtr();
SegmentIteratorPtr source = mappedSeg->getSourceIteratorPtr();
TopSegmentIteratorPtr top = std::dynamic_pointer_cast<TopSegmentIterator>(target);
TopSegmentIteratorPtr topCopy = top->clone();
do {
// FIXME: why isn't clone() polymorphic?
SegmentIteratorPtr newSource;
if (source->isTop()) {
newSource = std::dynamic_pointer_cast<TopSegmentIterator>(source)->clone();
} else {
newSource = std::dynamic_pointer_cast<BottomSegmentIterator>(source)->clone();
}
TopSegmentIteratorPtr newTop = topCopy->clone();
MappedSegmentPtr newMappedSeg(new MappedSegment(newSource, newTop));
assert(newMappedSeg->getGenome() == mappedSeg->getGenome());
assert(newMappedSeg->getSource()->getGenome() == mappedSeg->getSource()->getGenome());
results.push_back(newMappedSeg);
++added;
if (topCopy->tseg()->hasNextParalogy()) {
topCopy->toNextParalogy();
}
} while (topCopy->tseg()->hasNextParalogy() == true && topCopy->getLength() >= minLength &&
topCopy->getArrayIndex() != top->getArrayIndex());
} else if (mappedSeg->getGenome()->getParent() != NULL) {
hal_index_t rightCutoff = mappedSeg->getEndPosition();
BottomSegmentIteratorPtr bottom = mappedSeg->targetAsBottom();
hal_index_t startOffset = (hal_index_t)bottom->getStartOffset();
hal_index_t endOffset = (hal_index_t)bottom->getEndOffset();
TopSegmentIteratorPtr top = mappedSeg->getGenome()->getTopSegmentIterator();
top->toParseUp(bottom);
do {
TopSegmentIteratorPtr topNew = top->clone();
// we map the new target back to see how the offsets have
// changed. these changes are then applied to the source segment
// as deltas
BottomSegmentIteratorPtr bottomBack = bottom->clone();
bottomBack->toParseDown(topNew);
hal_index_t startBack = (hal_index_t)bottomBack->getStartOffset();
hal_index_t endBack = (hal_index_t)bottomBack->getEndOffset();
assert(startBack >= startOffset);
assert(endBack >= endOffset);
SegmentIteratorPtr newSource = mappedSeg->sourceClone();
hal_index_t startDelta = startBack - startOffset;
hal_index_t endDelta = endBack - endOffset;
assert((hal_index_t)newSource->getLength() > startDelta + endDelta);
newSource->slice(newSource->getStartOffset() + startDelta, newSource->getEndOffset() + endDelta);
MappedSegmentPtr newMappedSeg(new MappedSegment(newSource, topNew));
assert(newMappedSeg->isTop() == true);
assert(newMappedSeg->getSource()->getGenome() == mappedSeg->getSource()->getGenome());
added += mapSelf(newMappedSeg, results, minLength);
// stupid that we have to make this check but odn't want to
// make fundamental api change now
if (top->getEndPosition() != rightCutoff) {
top->toRight(rightCutoff);
} else {
break;
}
} while (true);
}
return added;
}
void MappedSegmentMapExtraParalogsTest::createCallBack(AlignmentPtr alignment)
{
vector<Sequence::Info> seqVec(1);
BottomSegmentIteratorPtr bi;
BottomSegmentStruct bs;
TopSegmentIteratorPtr ti;
TopSegmentStruct ts;
// Set up a case where all the segments of grandChild1 coalesce with
// the first segment of grandChild2, but only if using the root as
// the coalescence limit. Otherwise only the first segments map to
// each other.
Genome* root = alignment->addRootGenome("root");
Genome* parent = alignment->addLeafGenome("parent", "root", 1);
Genome* grandChild1 = alignment->addLeafGenome("grandChild1", "parent", 1);
Genome* grandChild2 = alignment->addLeafGenome("grandChild2", "parent", 1);
seqVec[0] = Sequence::Info("Sequence", 3, 0, 1);
root->setDimensions(seqVec);
seqVec[0] = Sequence::Info("Sequence", 9, 3, 3);
parent->setDimensions(seqVec);
seqVec[0] = Sequence::Info("Sequence", 9, 3, 0);
grandChild1->setDimensions(seqVec);
seqVec[0] = Sequence::Info("Sequence", 9, 3, 0);
grandChild2->setDimensions(seqVec);
root->setString("CCC");
parent->setString("CCCTACGTG");
grandChild1->setString("CCCTACGTG");
grandChild2->setString("CCCTACGTG");
bi = root->getBottomSegmentIterator();
bs.set(0, 3);
bs._children.push_back(pair<hal_size_t, bool>(0, false));
bs.applyTo(bi);
ti = parent->getTopSegmentIterator();
ts.set(0, 3, 0, false, NULL_INDEX, 1);
ts.applyTo(ti);
ti->toRight();
ts.set(3, 3, 0, false, NULL_INDEX, 2);
ts.applyTo(ti);
ti->toRight();
ts.set(6, 3, 0, false, NULL_INDEX, 0);
ts.applyTo(ti);
bi = parent->getBottomSegmentIterator();
bs.set(0, 3);
bs._children.clear();
bs._children.push_back(pair<hal_size_t, bool>(0, true));
bs._children.push_back(pair<hal_size_t, bool>(0, false));
bs.applyTo(bi);
bi->toRight();
bs.set(3, 3);
bs._children.clear();
bs._children.push_back(pair<hal_size_t, bool>(1, true));
bs._children.push_back(pair<hal_size_t, bool>(NULL_INDEX, true));
bs.applyTo(bi);
bi->toRight();
bs.set(6, 3);
bs._children.clear();
bs._children.push_back(pair<hal_size_t, bool>(2, true));
bs._children.push_back(pair<hal_size_t, bool>(NULL_INDEX, false));
bs.applyTo(bi);
ti = grandChild1->getTopSegmentIterator();
ts.set(0, 3, 0, true);
ts.applyTo(ti);
ti->toRight();
ts.set(3, 3, 1, true);
ts.applyTo(ti);
ti->toRight();
ts.set(6, 3, 2, true);
ts.applyTo(ti);
ti = grandChild2->getTopSegmentIterator();
ts.set(0, 3, 0, false);
ts.applyTo(ti);
ti->toRight();
ts.set(3, 3, NULL_INDEX, true);
ts.applyTo(ti);
ti->toRight();
ts.set(6, 3, NULL_INDEX, false);
ts.applyTo(ti);
parent->fixParseInfo();
}
void TopSegmentIteratorParseTest::createCallBack(AlignmentPtr alignment)
{
vector<Sequence::Info> seqVec(1);
BottomSegmentIteratorPtr bi;
BottomSegmentStruct bs;
TopSegmentIteratorPtr ti;
TopSegmentStruct ts;
// case 1: bottom segment aligns perfectly with top segment
Genome* case1 = alignment->addRootGenome("case1");
seqVec[0] = Sequence::Info("Sequence", 10, 2, 2);
case1->setDimensions(seqVec);
ti = case1->getTopSegmentIterator();
ts.set(0, 10, NULL_INDEX, false, 0, NULL_INDEX);
ts.applyTo(ti);
bi = case1->getBottomSegmentIterator();
bs.set(0, 10, 0);
bs.applyTo(bi);
// case 2: bottom segment is completely contained in top segment
Genome* case2 = alignment->addRootGenome("case2");
seqVec[0] = Sequence::Info("Sequence", 10, 2, 3);
case2->setDimensions(seqVec);
ti = case2->getTopSegmentIterator();
ts.set(0, 9, NULL_INDEX, false, 0, NULL_INDEX);
ts.applyTo(ti);
bi = case2->getBottomSegmentIterator();
bs.set(0, 3, 0);
bs.applyTo(bi);
bi->toRight();
bs.set(3, 4, 0);
bs.applyTo(bi);
bi->toRight();
bs.set(7, 3, 0);
bs.applyTo(bi);
// case 3 top segment is completely contained in bottom segment
Genome* case3 = alignment->addRootGenome("case3");
seqVec[0] = Sequence::Info("Sequence", 10, 3, 2);
case3->setDimensions(seqVec);
ti = case3->getTopSegmentIterator();
ts.set(0, 3, NULL_INDEX, false, 0);
ts.applyTo(ti);
ti->toRight();
ts.set(3, 4, NULL_INDEX, false, 0);
ts.applyTo(ti);
ti->toRight();
ts.set(7, 3, NULL_INDEX, false, 0);
ts.applyTo(ti);
bi = case3->getBottomSegmentIterator();
bs.set(0, 9, 0);
bs.applyTo(bi);
// case 4: top segment overhangs bottom segment on the left
Genome* case4 = alignment->addRootGenome("case4");
seqVec[0] = Sequence::Info("Sequence", 10, 2, 2);
case4->setDimensions(seqVec);
ti = case4->getTopSegmentIterator();
ts.set(0, 9, NULL_INDEX, false, 0);
ts.applyTo(ti);
bi = case4->getBottomSegmentIterator();
bs.set(0, 5, 0);
bs.applyTo(bi);
bi->toRight();
bs.set(5, 5, 0);
bs.applyTo(bi);
}