Python MinHash.add Examples

Example #1

def test_dna_mh(track_abundance):
    e1 = MinHash(n=5, ksize=4, track_abundance=track_abundance)
    e2 = MinHash(n=5, ksize=4, track_abundance=track_abundance)

    seq = 'ATGGCAGTGACGATGCCAG'
    e1.add_sequence(seq)
    for i in range(len(seq) - 3):
        e2.add(seq[i:i + 4])

    assert e1.get_mins() == e2.get_mins()
    print(e1.get_mins())
    assert 726311917625663847 in e1.get_mins()
    assert 3697418565283905118 in e1.get_mins()

Example #2

File: test_jaccard.py Project: dib-lab/sourmash

def test_dna_mh(track_abundance):
    e1 = MinHash(n=5, ksize=4, track_abundance=track_abundance)
    e2 = MinHash(n=5, ksize=4, track_abundance=track_abundance)

    seq = 'ATGGCAGTGACGATGCCAG'
    e1.add_sequence(seq)
    for i in range(len(seq) - 3):
        e2.add(seq[i:i + 4])

    assert e1.get_mins() == e2.get_mins()
    print(e1.get_mins())
    assert 726311917625663847 in e1.get_mins()
    assert 3697418565283905118 in e1.get_mins()

Example #3

File: test_jaccard.py Project: dib-lab/sourmash

def test_protein_mh(track_abundance):
    e1 = MinHash(n=5, ksize=6, is_protein=True,
                    track_abundance=track_abundance)
    e2 = MinHash(n=5, ksize=6, is_protein=True,
                    track_abundance=track_abundance)

    seq = 'ATGGCAGTGACGATGCCG'
    e1.add_sequence(seq)

    for i in range(len(seq) - 5):
        kmer = seq[i:i + 6]
        e2.add(kmer)

    assert e1.get_mins() == e2.get_mins()
    assert 901193879228338100 in e1.get_mins()

Example #4

File: test_jaccard.py Project: xmnlab/sourmash

def test_protein_mh(track_abundance):
    e1 = MinHash(n=5, ksize=6, is_protein=True,
                    track_abundance=track_abundance)
    e2 = MinHash(n=5, ksize=6, is_protein=True,
                    track_abundance=track_abundance)

    seq = 'ATGGCAGTGACGATGCCG'
    e1.add_sequence(seq)

    for i in range(len(seq) - 5):
        kmer = seq[i:i + 6]
        e2.add(kmer)

    assert e1.get_mins() == e2.get_mins()
    assert 901193879228338100 in e1.get_mins()