C++ BottomSegmentIteratorPtr::getStartPosition方法代码示例

void TopSegmentIteratorReverseTest::checkCallBack(const Alignment *alignment) {
    BottomSegmentIteratorPtr bi;
    TopSegmentIteratorPtr ti, ti2;

    const Genome *parent1 = alignment->openGenome("parent1");
    const Genome *child1 = alignment->openGenome("child1");

    ti = child1->getTopSegmentIterator();
    bi = parent1->getBottomSegmentIterator();

    ti2 = child1->getTopSegmentIterator();
    ti2->toChild(bi, 0);

    CuAssertTrue(_testCase, ti->getStartPosition() == 0);
    CuAssertTrue(_testCase, ti->getLength() == 10);
    CuAssertTrue(_testCase, ti->getReversed() == false);

    CuAssertTrue(_testCase, ti2->getStartPosition() == 9);
    CuAssertTrue(_testCase, ti2->getLength() == 10);
    CuAssertTrue(_testCase, ti2->getReversed() == true);

    bi->slice(1, 3);
    ti2->toChild(bi, 0);

    CuAssertTrue(_testCase, bi->getStartPosition() == 1);
    CuAssertTrue(_testCase, bi->getLength() == 6);
    CuAssertTrue(_testCase, ti2->getStartPosition() == 8);
    CuAssertTrue(_testCase, ti2->getLength() == 6);

    string buffer;
    bi->getString(buffer);
    CuAssertTrue(_testCase, buffer == "CCTACG");
    ti2->getString(buffer);
    CuAssertTrue(_testCase, buffer == "CACGTA");

    bi = child1->getBottomSegmentIterator();
    CuAssertTrue(_testCase, bi->getReversed() == false);

    ti->toParseUp(bi);
    CuAssertTrue(_testCase, ti->getStartPosition() == 0);
    CuAssertTrue(_testCase, ti->getLength() == 5);

    bi->toReverse();
    ti->toParseUp(bi);
    CuAssertTrue(_testCase, ti->getStartPosition() == 4);
    CuAssertTrue(_testCase, ti->getLength() == 5);

    bi->toReverse();
    CuAssertTrue(_testCase, bi->getReversed() == false);
    bi->toRight();
    ti->toParseUp(bi);
    CuAssertTrue(_testCase, ti->getStartPosition() == 5);
    CuAssertTrue(_testCase, ti->getLength() == 5);

    bi->toReverse();
    ti->toParseUp(bi);
    CuAssertTrue(_testCase, ti->getStartPosition() == 9);
    CuAssertTrue(_testCase, ti->getLength() == 5);
}

void checkBottomSegments(Genome *genome, hal_size_t width, CuTest *testCase) {
    hal_size_t numChildren = genome->getNumChildren();
    BottomSegmentIteratorPtr bottomIt = genome->getBottomSegmentIterator();
    hal_size_t n = genome->getNumBottomSegments();
    hal_index_t startPos = 0;
    for (; bottomIt->getArrayIndex() < n; bottomIt->toRight(), startPos += width) {
        CuAssertTrue(testCase, bottomIt->getStartPosition() == startPos);

        for (hal_size_t i = 0; i < numChildren; i++) {
            if (startPos < 170) {
                CuAssertStrEquals(testCase, "Sequence2", genome->getSequenceBySite(startPos)->getName().c_str());
                CuAssertTrue(testCase, bottomIt->bseg()->getChildIndex(i) == 13 + bottomIt->getArrayIndex());
            } else {
                CuAssertStrEquals(testCase, "Sequence1", genome->getSequenceBySite(startPos)->getName().c_str());
                CuAssertTrue(testCase, bottomIt->bseg()->getChildIndex(i) == bottomIt->getArrayIndex() - 17);
            }
        }
    }
}

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 Genome::copyBottomSegments(Genome *dest) const
{
  assert(getNumBottomSegments() == dest->getNumBottomSegments());
  hal_size_t inNc = getNumChildren();
  hal_size_t outNc = dest->getNumChildren();
  // The child indices aren't consistent across files--make sure each bottom
  // segment points to the correct children
  vector<string> inChildNames;
  vector<string> outChildNames;
  for (hal_size_t inChild = 0; inChild < inNc; ++inChild)
  {
    inChildNames.push_back(getChild(inChild)->getName());
  }
  for (hal_size_t outChild = 0; outChild < outNc; ++outChild)
  {
    outChildNames.push_back(dest->getChild(outChild)->getName());
  }
  map<hal_size_t, hal_size_t> inChildToOutChild;
  for (hal_size_t inChild = 0; inChild < inNc; inChild++)
  {
    hal_size_t outChild;
    for (outChild = 0; outChild < outNc; outChild++)
    {
      if (inChildNames[inChild] == outChildNames[outChild])
      {
        inChildToOutChild[inChild] = outChild;
        break;
      }
    }
    if (outChild == outNc)
    {
      inChildToOutChild[inChild] = outNc;
    }
  }

  // Go through each sequence in this genome, find the matching
  // sequence in the dest genome, then copy over the segments for each
  // sequence.
  SequenceIteratorConstPtr seqIt = getSequenceIterator();
  SequenceIteratorConstPtr seqEndIt = getSequenceEndIterator();

  for (; seqIt != seqEndIt; seqIt->toNext())
  {
    const Sequence *inSeq = seqIt->getSequence();
    const Sequence *outSeq = dest->getSequence(inSeq->getName());
    BottomSegmentIteratorPtr inBot = inSeq->getBottomSegmentIterator();
    BottomSegmentIteratorPtr outBot = outSeq->getBottomSegmentIterator();

    cout << "DEBUG: inSeq name: " << inSeq->getName() << ", outSeq name: " << outSeq->getName() << endl;

    if (inSeq->getName() != outSeq->getName()) {
      // This check is important enough that it can't be an assert.
      stringstream ss;
      ss << "When copying bottom segments: segment #" << inBot->getArrayIndex() << " of source genome is from sequence " << inBot->getSequence()->getName() << ", but segment #" << outBot->getArrayIndex() << " is from sequence " << outBot->getSequence()->getName();
      throw hal_exception(ss.str());
    }

    if (inSeq->getNumBottomSegments() != outSeq->getNumBottomSegments()) {
      stringstream ss;
      ss << "When copying bottom segments: sequence " << inSeq->getName() << " has " << inSeq->getNumBottomSegments() << " in genome " << getName() << ", while it has " << outSeq->getNumBottomSegments() << " in genome " << dest->getName();
      throw hal_exception(ss.str());      
    }

    hal_index_t inSegmentEnd = inSeq->getBottomSegmentArrayIndex() + inSeq->getNumBottomSegments();
    cout << "DEBUG: inSegmentStart: " << inSeq->getBottomSegmentArrayIndex() << " inSegmentEnd: " << inSegmentEnd << " num bottom segments: " << inSeq->getNumBottomSegments() << endl;
    for (; inBot->getArrayIndex() < inSegmentEnd; inBot->toRight(),
           outBot->toRight())
    {
      hal_index_t outStartPosition = inBot->getStartPosition() - inSeq->getStartPosition() + outSeq->getStartPosition();

      cout << "Decided on outStartPosition " << outStartPosition << " for seg index " << outBot->getArrayIndex() << " (src index " << inBot->getArrayIndex() << ")" << endl;

      if (dest->getSequenceBySite(outStartPosition) != outSeq) {
        stringstream ss;
        ss << "When copying bottom segments from " << getName() << " to " << dest->getName() << ": expected destination sequence " << outSeq->getName() << " for segment # " << inBot->getArrayIndex() << " but got " << dest->getSequenceBySite(outStartPosition)->getName();
        throw hal_exception(ss.str());
      }
      outBot->setCoordinates(outStartPosition, inBot->getLength());
      for(hal_size_t inChild = 0; inChild < inNc; inChild++) {
        hal_size_t outChild = inChildToOutChild[inChild];
        if (outChild != outNc) {
          outBot->setChildIndex(outChild, inBot->getChildIndex(inChild));
          cout << "genome " << getName() << ": Set child index " << inChild << " to " << inBot->getChildIndex(inChild) << endl;
          outBot->setChildReversed(outChild, inBot->getChildReversed(inChild));
        }
      }
      outBot->setTopParseIndex(inBot->getTopParseIndex());
    }
  }
}