def __init__(self, extend_length=50, overlap_length=None, reads1=None, reads2=None, seq=None, ext_min_cov=5, ext_min_ratio=2, verbose=0, seed_length=None, seed_min_count=10, seed_max_count=100000000, kmc_threads=1, map_threads=1, sequences_to_ignore=None, contigs_to_check=None):
if contigs_to_check is None:
contigs_to_check = {}
if sequences_to_ignore is None:
sequences_to_ignore = set()
self.verbose = verbose
self.kmc_threads = kmc_threads
self.map_threads = map_threads
self.extend_length = extend_length
self.ext_min_cov = ext_min_cov
self.ext_min_ratio = ext_min_ratio
self.seed_lengths = []
self.overlap_length = overlap_length
if seq is None:
if reads1 is None:
raise Error('Cannot construct Seed object. Need reads when no seq has been given')
kmer_counts = kcount.get_most_common_kmers(reads1, reads2, most_common=1, min_count=seed_min_count, max_count=seed_max_count, kmer_length=seed_length, verbose=self.verbose, ignore_seqs=sequences_to_ignore, contigs_to_check=contigs_to_check, kmc_threads=self.kmc_threads, map_threads=self.map_threads)
if len(kmer_counts) == 1:
self.seq = list(kmer_counts.keys())[0]
if self.verbose:
print('Made new seed. kmer coverage', list(kmer_counts.values())[0], 'and seed is', self.seq, flush=True)
else:
self.seq = None
else:
self.seq = seq
if self.seq is not None:
if overlap_length is None:
self.overlap_length = len(self.seq)
else:
self.overlap_length = overlap_length
else:
self.overlap_length = None
def test_get_most_common_kmers(self):
'''Test get_most_common_kmers'''
reads1 = os.path.join(data_dir, 'kcount_test.get_commonest_kmer_1.fa')
reads2 = os.path.join(data_dir, 'kcount_test.get_commonest_kmer_2.fa')
counts = kcount.get_most_common_kmers(reads1,
reads2,
kmer_length=10,
head=100000,
min_count=2,
max_count=4,
most_common=100,
method='kmc')
self.assertDictEqual({'AGCTAAAACT': 2, 'CTATATCTCA': 3}, counts)
def __init__(self,
extend_length=50,
overlap_length=None,
reads1=None,
reads2=None,
seq=None,
ext_min_cov=5,
ext_min_ratio=2,
verbose=0,
seed_length=None,
seed_min_count=10,
seed_max_count=100000000,
kmc_threads=1,
map_threads=1,
sequences_to_ignore=None,
contigs_to_check=None):
if contigs_to_check is None:
contigs_to_check = {}
if sequences_to_ignore is None:
sequences_to_ignore = set()
self.verbose = verbose
self.kmc_threads = kmc_threads
self.map_threads = map_threads
self.extend_length = extend_length
self.ext_min_cov = ext_min_cov
self.ext_min_ratio = ext_min_ratio
self.seed_lengths = []
self.overlap_length = overlap_length
if seq is None:
if reads1 is None:
raise Error(
'Cannot construct Seed object. Need reads when no seq has been given'
)
kmer_counts = kcount.get_most_common_kmers(
reads1,
reads2,
most_common=1,
min_count=seed_min_count,
max_count=seed_max_count,
kmer_length=seed_length,
verbose=self.verbose,
ignore_seqs=sequences_to_ignore,
contigs_to_check=contigs_to_check,
kmc_threads=self.kmc_threads,
map_threads=self.map_threads)
if len(kmer_counts) == 1:
self.seq = list(kmer_counts.keys())[0]
if self.verbose:
print('Made new seed. kmer coverage',
list(kmer_counts.values())[0],
'and seed is',
self.seq,
flush=True)
else:
self.seq = None
else:
self.seq = seq
if self.seq is not None:
if overlap_length is None:
self.overlap_length = len(self.seq)
else:
self.overlap_length = overlap_length
else:
self.overlap_length = None
def test_get_most_common_kmers(self):
'''Test get_most_common_kmers'''
reads1 = os.path.join(data_dir, 'kcount_test.get_commonest_kmer_1.fa')
reads2 = os.path.join(data_dir, 'kcount_test.get_commonest_kmer_2.fa')
counts = kcount.get_most_common_kmers(reads1, reads2, kmer_length=10, head=100000, min_count=2, max_count=4, most_common=100, method='kmc')
self.assertDictEqual({'AGCTAAAACT': 2, 'CTATATCTCA': 3}, counts)