Method kmerBased

src/rnabloom/util/SeqSubsampler.java:120–337

(ArrayList<? extends BitSequence> seqs,
            String outSubsampleFasta,
            long bfSize, int k, int numHash, boolean stranded, 
            int maxMultiplicity, int maxEdgeClip, boolean verbose,
            int numThreads)

    }
        
    public static void kmerBased(ArrayList<? extends BitSequence> seqs,
            String outSubsampleFasta,
            long bfSize, int k, int numHash, boolean stranded, 
            int maxMultiplicity, int maxEdgeClip, boolean verbose,
            int numThreads) throws IOException, InterruptedException, ExecutionException {
        
        System.out.println("k-mers: n=2, k=" + k);
        int numSeq = seqs.size();
        
        final int shift = k + 1;
        final int shiftD = k;
        final int shiftI = k + 2;
        final int missingChainThreshold = k + shift;
        int numSubsample = 0;
        float fpr;

        ConcurrentLinkedQueue<String> subQueue = new ConcurrentLinkedQueue<>();
        FastaWriter subWriter  = new FastaWriter(outSubsampleFasta, false);
        FastaWriterWorker subWriterWorker = new FastaWriterWorker(subQueue, subWriter, "s");
        Thread subWriterThread = new Thread(subWriterWorker);
        subWriterThread.start();
        
        CountingBloomFilter cbf;
        ForkJoinPool customThreadPool = new ForkJoinPool(numThreads);
        
        if (stranded) {
            HashFunction h = new HashFunction(k);
            NTHashIterator hashItr = h.getHashIterator(1);
            cbf = new CountingBloomFilter(bfSize, numHash, h);
            
            for (int seqIndex=0; seqIndex<numSeq; ++seqIndex) {
                String seq = seqs.get(seqIndex).toString();

                if (hashItr.start(seq)) {
                    int seqLen = seq.length();
                    boolean tooShort = seqLen < 3 * maxEdgeClip;
                    int missingChainLen = 0;
                    boolean write = false;
                    int numKmers = seqLen - k + 1;

                    // get hash value of all kmers
                    long[] kmerHashVals = new long[numKmers];
                    for (int i=0; i<numKmers; ++i) {
                        hashItr.next();
                        kmerHashVals[i] = hashItr.hVals[0];
                    }

                    final int start, end;
                    if (tooShort) {
                        start = 0;
                        end = numKmers - shift;
                    }
                    else {
                        start = maxEdgeClip;
                        end = numKmers - maxEdgeClip - shift;
                    }

                    // look up counts of k-mer pairs