const std::vector<std::vector<unsigned > > SeedMemoryManager<KmerT>::getNotFoundMatchesCount(
const flowcell::TileMetadataList &unprocessedTiles,
const flowcell::BarcodeMetadataList &barcodeMetadataList,
const ReadMetadataList &readMetadataList,
const matchFinder::TileClusterInfo &foundMatches) const
{
ISAAC_ASSERT_MSG(unprocessedTiles.front().getIndex() <= unprocessedTiles.back().getIndex(),
"Expected tiles ordered by index");
std::vector<std::vector<unsigned > > ret(readMetadataList.size(), std::vector<unsigned>(
unprocessedTiles.back().getIndex() + 1));
BOOST_FOREACH(const flowcell::ReadMetadata &readMetadata, readMetadataList_)
{
const unsigned readIndex = readMetadata.getIndex();
BOOST_FOREACH(const flowcell::TileMetadata &tileMetadata, unprocessedTiles)
{
const unsigned tileIndex = tileMetadata.getIndex();
const std::vector<matchFinder::ClusterInfo> &oneTileInfo = foundMatches.at(tileIndex);
// match only clusters where no matches were found so far
ISAAC_ASSERT_MSG(oneTileInfo.size() == tileMetadata.getClusterCount(), "allTiles and foundMatches geometries must match");
ret.at(readIndex).at(tileIndex) =
std::count_if(oneTileInfo.begin(), oneTileInfo.end(),
boost::bind(&willLoadSeeds,
boost::ref(barcodeMetadataList), _1, readIndex));
}
}
return ret;
}
inline unsigned getLaneNumber(const flowcell::TileMetadataList& tiles)
{
ISAAC_ASSERT_MSG(
tiles.end()
== std::find_if(
tiles.begin(),
tiles.end(),
boost::bind(&flowcell::TileMetadata::getLane, _1)
!= tiles.front().getLane()),
"Expected all tiles to belong to the same lane");
return tiles.front().getLane();
}
MatchFinder<KmerT>::MatchFinder(
const reference::SortedReferenceMetadataList &sortedReferenceList,
const bfs::path & tempDirectory,
const flowcell::TileMetadataList &tiles,
const flowcell::ReadMetadataList &readMetadataList,
const SeedMetadataList &seedMetadataList,
const unsigned iteration,
const bool ignoreNeighbors,
const bool ignoreRepeats,
const unsigned repeatThreshold,
const unsigned neighborhoodSizeThreshold,
MatchTally &matchTally,
matchFinder::TileClusterInfo &foundMatches,
common::ThreadVector &threads,
const unsigned coresMax,
const unsigned tempSaversMax,
const unsigned unavailableFileHandles)
: kmerSourceMetadataList_(getMaskFilesList(sortedReferenceList))
, referenceContigKaryotypes_(getReferenceContigKaryotypes(sortedReferenceList))
, seedMetadataList_(seedMetadataList)
, iteration_(iteration)
, ignoreNeighbors_(ignoreNeighbors)
, ignoreRepeats_(ignoreRepeats)
, repeatThreshold_(repeatThreshold)
, neighborhoodSizeThreshold_(neighborhoodSizeThreshold)
, foundExactMatchesOnly_(foundMatches)
, threads_(threads)
, threadsMax_(coresMax)
, maxTilesAtATime_(verifyMaxTileCount(unavailableFileHandles, tempSaversMax, tiles.size()))
, threadRepeatLists_(threadsMax_, std::vector<ReferenceKmer>(repeatThreshold_ + 1))
, threadNeighborsLists_(threadsMax_, std::vector<ReferenceKmer>(neighborhoodSizeThreshold_ + 1))
, threadMatchDistributions_(threadsMax_, MatchDistribution(sortedReferenceList))
, matchWriter_(matchTally, maxTilesAtATime_, tiles.back().getIndex())
, threadReferenceFileBuffers_(
threadsMax_,
io::FileBufCache<io::FileBufWithReopen>(1, std::ios_base::binary|std::ios_base::in,
std::max_element(kmerSourceMetadataList_.begin(), kmerSourceMetadataList_.end(),
boost::bind(&KmerSourceMetadata::getPathSize, _1)<
boost::bind(&KmerSourceMetadata::getPathSize, _2))->getPathSize()))
{
ISAAC_THREAD_CERR << "Constructing the match finder" << std::endl;
ISAAC_THREAD_CERR << "Constructing the match finder done" << std::endl;
}
void ParallelSeedLoader<ReaderT, KmerT>::loadSeeds(
const flowcell::TileMetadataList &tiles,
const matchFinder::TileClusterInfo &tileClusterBarcode,
std::vector<Seed<KmerT> > &seeds,
common::ScopedMallocBlock &mallocBlock)
{
// Start and execute the threads
ISAAC_THREAD_CERR << "Loading data on " << inputLoadersMax_ << " threads" << std::endl;
std::vector<flowcell::TileMetadata>::const_iterator nextTile = tiles.begin();
BaseT::reset(tiles, seeds, tileClusterBarcode);
threads_.execute(boost::bind(&ParallelSeedLoader::load, this,
boost::ref(tileClusterBarcode),
boost::ref(nextTile),
tiles.end(), _1),
inputLoadersMax_);
BaseT::sortSeeds(seeds, mallocBlock, threads_, coresMax_);
}