void TopSegmentIteratorToSiteTest::checkGenome(const Genome *genome) {
TopSegmentIteratorPtr ti = genome->getTopSegmentIterator();
for (hal_index_t pos = 0; pos < (hal_index_t)genome->getSequenceLength(); ++pos) {
ti->toSite(pos);
CuAssertTrue(_testCase, ti->getStartPosition() == pos);
CuAssertTrue(_testCase, ti->getLength() == 1);
ti->toSite(pos, false);
CuAssertTrue(_testCase, pos >= ti->getStartPosition() && pos < ti->getStartPosition() + (hal_index_t)ti->getLength());
CuAssertTrue(_testCase, ti->getLength() == ti->getTopSegment()->getLength());
}
}
void TopSegmentIteratorParseTest::checkCallBack(const Alignment *alignment) {
BottomSegmentIteratorPtr bi;
TopSegmentIteratorPtr ti;
// case 1
const Genome *case1 = alignment->openGenome("case1");
ti = case1->getTopSegmentIterator();
bi = case1->getBottomSegmentIterator();
ti->toParseUp(bi);
CuAssertTrue(_testCase, bi->getStartPosition() == ti->getStartPosition());
CuAssertTrue(_testCase, bi->getLength() == ti->getLength());
bi->slice(3, 1);
ti->toParseUp(bi);
CuAssertTrue(_testCase, bi->getLength() == bi->getBottomSegment()->getLength() - 4);
CuAssertTrue(_testCase, bi->getStartPosition() == ti->getStartPosition());
CuAssertTrue(_testCase, bi->getLength() == ti->getLength());
// case 2
const Genome *case2 = alignment->openGenome("case2");
ti = case2->getTopSegmentIterator();
bi = case2->getBottomSegmentIterator(1);
ti->toParseUp(bi);
CuAssertTrue(_testCase, bi->getStartPosition() == ti->getStartPosition());
bi->slice(1, 1);
ti->toParseUp(bi);
CuAssertTrue(_testCase, bi->getStartPosition() == ti->getStartPosition());
// case 3
const Genome *case3 = alignment->openGenome("case3");
ti = case3->getTopSegmentIterator();
bi = case3->getBottomSegmentIterator();
ti->toParseUp(bi);
CuAssertTrue(_testCase, bi->getStartPosition() == ti->getStartPosition());
bi->slice(2, 1);
ti->toParseUp(bi);
CuAssertTrue(_testCase, bi->getStartPosition() == ti->getStartPosition());
// case 4
const Genome *case4 = alignment->openGenome("case4");
ti = case4->getTopSegmentIterator();
bi = case4->getBottomSegmentIterator(1);
ti->toParseUp(bi);
CuAssertTrue(_testCase, bi->getStartPosition() == ti->getStartPosition());
bi->slice(2, 2);
ti->toParseUp(bi);
CuAssertTrue(_testCase, bi->getStartPosition() == ti->getStartPosition());
}
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;
}
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 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());
}
// 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;
}