void ExtractNeighborsWorkflow::run()
{
const reference::SortedReferenceMetadata::MaskFiles &maskFiles =
xml_.getMaskFileList(oligo::KmerTraits<KmerT>::KMER_BASES);
if (maskFiles.empty())
{
BOOST_THROW_EXCEPTION(isaac::common::PreConditionException("No mask files in " + sortedReferenceMetadata_.string()));
}
const reference::SortedReferenceMetadata::Contigs contigs = xml_.getKaryotypeOrderedContigs();
const std::vector<unsigned long> contigOffsets = reference::computeContigOffsets(contigs);
std::vector<bool> neighbors(reference::genomeLength(contigs), false);
std::vector<bool> highRepeats(highRepeatsFilePath_.empty() ? 0 : reference::genomeLength(contigs), true);
// there could be mutliple mask widths in the xml. Just pick one.
const unsigned maskWidth = xml_.getDefaultMaskWidth();
BOOST_FOREACH(const reference::SortedReferenceMetadata::MaskFile &maskFile, maskFiles)
{
//Don't reprocess redundant mask files of different widths
if (maskWidth == maskFile.maskWidth)
{
scanMaskFile<KmerT>(maskFile, contigOffsets, neighbors, highRepeats);
}
}
dumpResults(neighbors, highRepeats);
}
void ExtractNeighborsWorkflow::run()
{
const reference::SortedReferenceMetadata::MaskFiles &maskFiles =
xml_.getMaskFileList(oligo::KmerTraits<KmerT>::KMER_BASES);
if (maskFiles.empty())
{
BOOST_THROW_EXCEPTION(isaac::common::PreConditionException("No mask files in " + sortedReferenceMetadata_.string()));
}
const reference::SortedReferenceMetadata::Contigs contigs = xml_.getContigs();
const std::vector<uint64_t> contigOffsets = reference::computeContigOffsets(contigs);
std::vector<unsigned> karyotypes;
karyotypes.reserve(xml_.getContigs().size());
std::transform(xml_.getContigs().begin(), xml_.getContigs().end(), std::back_inserter(karyotypes),
boost::bind(&reference::SortedReferenceMetadata::Contig::karyotypeIndex_, _1));
std::vector<bool> neighbors(reference::genomeLength(contigs), false);
std::vector<bool> highRepeats(highRepeatsFilePath_.empty() ? 0 : reference::genomeLength(contigs), true);
// there could be multiple mask widths in the xml. Just fail if there are.
unsigned maskWidth = -1U;
BOOST_FOREACH(const reference::SortedReferenceMetadata::MaskFile &maskFile, maskFiles)
{
if (-1U == maskWidth)
{
maskWidth = maskFile.maskWidth;
}
ISAAC_ASSERT_MSG(maskWidth == maskFile.maskWidth, "Mixed mask widths are not supported");
scanMaskFile<KmerT>(maskFile, contigOffsets, karyotypes, neighbors, highRepeats);
}
dumpResults(neighbors, highRepeats);
}