Examples of org.broadinstitute.gatk.utils.collections.CountSet

Package org.broadinstitute.gatk.utils.collections

Examples of org.broadinstitute.gatk.utils.collections.CountSet

org.broadinstitute.gatk.utils.collections.CountSet
Efficient implementation for a small set of integer primitive values.
It includes a increment operation incAll which is convenient when analyzing the read-threading graphs. Nevertheless it can be also be used in general purpose.

It does not provide a O(1) look-up of its elements though. These are kept in a sorted array so look up is implemented using a binary search O(log n). Therefore it might not be optimal for problems that require large integer sets.

Also note that addition can be costly for large sets unless done in order: O(n).
@author Valentin Ruano-Rubio <valentin@broadinstitute.org>

        final Set<ReadSegmentCost> readSegmentCosts = new TreeSet<>(ReadSegmentComparator.INSTANCE);


        final int readStart = beforeBlockReadOffset + kmerSize;
        final int readEnd = Math.max(readStart, afterBlockReadOffset + kmerSize - 1);
        final byte[][] pathBases = new byte[acrossBlockPaths.size()][];
        final CountSet pathSizes = new CountSet(acrossBlockPaths.size());
        int nextPath = 0;


        // Complete the read segment cost with the corresponding path bases
        for (final Route<MultiDeBruijnVertex, MultiSampleEdge> p : acrossBlockPaths) {
            final ReadSegmentCost readSegmentCost = new ReadSegmentCost(read, p, 0);
            pathBases[nextPath++] = readSegmentCost.bases = eventBlockPathBases(p, includePathEnds);
            pathSizes.add(readSegmentCost.bases.length);
            readSegmentCosts.add(readSegmentCost);
        }


        // Add the read 'path size'.
        pathSizes.add(readEnd - readStart);


        final byte[] readBases = hmm.getReadBases();


        // Perform right clipping of bases that are common to all paths and read.
        int rightClipping = !doClipping ? 0 : calculateRightClipping(readEnd, pathBases, readBases,pathSizes);

View Full Code Here

     */
    public AssemblyResultSet() {
        assemblyResultByKmerSize = new LinkedHashMap<>(4);
        haplotypes = new LinkedHashSet<>(10);
        assemblyResultByHaplotype = new LinkedHashMap<>(10);
        kmerSizes = new CountSet(4);
    }

View Full Code Here

        int index = currentVertex.getSequence().length - 1;
        nonAnchorableDueToRepeats[index] = !sourceIsAnchorable;




        // We keep record on all alternative path lengths:
        final CountSet pathLengths = new CountSet(haplotypes.size());
        pathLengths.setTo(0);


        // Now we go through the reference path and determine which vertices are not part of event block.
        // We keep track of open divergent paths in expectedRejoins. Thus only those vertices traversed
        // when exptectedRejoins size 0 can be anchorable:
        while (currentVertex != null) {

View Full Code Here

        depths.add(prefixSizes);


        final Set<MultiDeBruijnVertex> visited = new HashSet<>();
        if (!exhaustive) visited.add(startVertex);
        while (!queue.isEmpty()) {
            final CountSet depth = depths.remove();
            final MultiDeBruijnVertex v = queue.remove();
            if (referenceVertices.contains(v)) {
                if (referenceWithinBoundaries.contains(v)) {
                    final CountSet previous = expectedRejoins.get(v);
                    if (previous == null)
                        expectedRejoins.put(v, depth.clone());
                    else
                        previous.addAll(depth);
                }
            } else {
                final CountSet depthPlusOne = depth.clone();
                depthPlusOne.incAll(1);
                final Set<MultiSampleEdge> nextEdges = backwards ? incomingEdgesOf(v) : outgoingEdgesOf(v);
                for (final MultiSampleEdge e : nextEdges) {
                    final MultiDeBruijnVertex w = backwards ? getEdgeSource(e) : getEdgeTarget(e);
                    if (visited.contains(w)) // avoid repetitive work.
                        continue;

View Full Code Here

            final ReadAnchoring anchoring, final boolean backwards,
            final MultiDeBruijnVertex leftVertex, final MultiDeBruijnVertex rightVertex) {


        MultiDeBruijnVertex currentVertex = backwards ? rightVertex : leftVertex;
        boolean foundEvent = false;
        final CountSet pathSizes = new CountSet(10); // typically more than enough.
        pathSizes.setTo(0);


        // Map between reference vertices where there is some expected open alternative path rejoining and the
        // predicted length of paths rejoining at that point counting from the beginning of the block.
        final Map<MultiDeBruijnVertex, CountSet> expectedAlternativePathRejoins = new HashMap<>(4);


        // Keeps record of possible left-clipping veritces; those that are located before any event path furcation
        // has been found. The value indicates the blockLength at the time we traverse that node.
        final Deque<Pair<MultiDeBruijnVertex, Integer>> possibleClippingPoints = new LinkedList<>();


        // We keep the distance from the beggining of the block (leftVertex).
        int blockLength = 0;
        while (currentVertex != null) {
            int openingDegree = backwards ? graph.outDegreeOf(currentVertex) : graph.inDegreeOf(currentVertex);
            if (openingDegree > 1) {
                final CountSet joiningPathLengths = expectedAlternativePathRejoins.remove(currentVertex);
                if (joiningPathLengths != null)
                    pathSizes.addAll(joiningPathLengths);
            }
            final boolean isValidBlockEnd = isValidBlockEnd(anchoring, currentVertex, expectedAlternativePathRejoins);
            if (foundEvent && isValidBlockEnd) // !gotcha we found a valid block end.

View Full Code Here

        for (final MultiSampleEdge e : nextEdges) {
            final MultiDeBruijnVertex nextVertex = backwards ? graph.getEdgeSource(e) : graph.getEdgeTarget(e);
            if (e.isRef())
                nextReferenceVertex = nextVertex;
            else {
                final CountSet pathSizesPlusOne = pathSizes.clone();
                pathSizesPlusOne.incAll(1);
                graph.calculateRejoins(nextVertex, expectedAlternativePathRejoins, anchoring.referenceWithinAnchorsMap.keySet(), pathSizesPlusOne, true, backwards);
            }
        }
        return nextReferenceVertex;
    }

View Full Code Here

            final ReadAnchoring anchoring, final boolean backwards,
            final MultiDeBruijnVertex leftVertex, final MultiDeBruijnVertex rightVertex) {


        MultiDeBruijnVertex currentVertex = backwards ? rightVertex : leftVertex;
        boolean foundEvent = false;
        final CountSet pathSizes = new CountSet(10); // typically more than enough.
        pathSizes.setTo(0);


        // Map between reference vertices where there is some expected open alternative path rejoining and the
        // predicted length of paths rejoining at that point counting from the beginning of the block.
        final Map<MultiDeBruijnVertex, CountSet> expectedAlternativePathRejoins = new HashMap<>(4);


        // Keeps record of possible left-clipping veritces; those that are located before any event path furcation
        // has been found. The value indicates the blockLength at the time we traverse that node.
        final Deque<Pair<MultiDeBruijnVertex, Integer>> possibleClippingPoints = new LinkedList<>();


        // We keep the distance from the beggining of the block (leftVertex).
        int blockLength = 0;
        while (currentVertex != null) {
            int openingDegree = backwards ? graph.outDegreeOf(currentVertex) : graph.inDegreeOf(currentVertex);
            if (openingDegree > 1) {
                final CountSet joiningPathLengths = expectedAlternativePathRejoins.remove(currentVertex);
                if (joiningPathLengths != null)
                    pathSizes.addAll(joiningPathLengths);
            }
            final boolean isValidBlockEnd = isValidBlockEnd(anchoring, currentVertex, expectedAlternativePathRejoins);
            if (foundEvent && isValidBlockEnd) // !gotcha we found a valid block end.

View Full Code Here

        for (final MultiSampleEdge e : nextEdges) {
            final MultiDeBruijnVertex nextVertex = backwards ? graph.getEdgeSource(e) : graph.getEdgeTarget(e);
            if (e.isRef())
                nextReferenceVertex = nextVertex;
            else {
                final CountSet pathSizesPlusOne = pathSizes.clone();
                pathSizesPlusOne.incAll(1);
                graph.calculateRejoins(nextVertex, expectedAlternativePathRejoins, anchoring.referenceWithinAnchorsMap.keySet(), pathSizesPlusOne, true, backwards);
            }
        }
        return nextReferenceVertex;
    }

View Full Code Here

TOP

Related Classes of org.broadinstitute.gatk.utils.collections.CountSet

org.broadinstitute.gatk.tools.walkers.haplotypecaller.AssemblyResultSet

org.broadinstitute.gatk.tools.walkers.haplotypecaller.EventBlockFinder

org.broadinstitute.gatk.tools.walkers.haplotypecaller.GraphBasedLikelihoodCalculationEngineInstance

org.broadinstitute.gatk.tools.walkers.haplotypecaller.readthreading.HaplotypeGraph

All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.