def test_odd(self):
seqs = [
SeqRecord(Seq('ATGACCAACTGA'), id='1'),
SeqRecord(Seq('ATGACCAACTGT'), id='2'),
SeqRecord(Seq('ATGACCAACTGC'), id='3'),
SeqRecord(Seq('ATGACCAACTGG'), id='4'),
SeqRecord(Seq('ATGACCAACTAT'), id='5'),
SeqRecord(Seq('ATGACCAACTAC'), id='6'),
SeqRecord(Seq('TTGACCAACTAC'), id='7'),
]
batch_iter = sequtils.seq_batch_iter(seqs, 3)
self.assertIsInstance(batch_iter, types.GeneratorType)
batch_iter = list(batch_iter)
self.assertEqual(len(batch_iter), 3)
new_seqs = []
for seq_iter in batch_iter:
self.assertIsInstance(seq_iter, seqsift.utils.dataio.BufferedIter)
for s in seq_iter:
new_seqs.append(s)
self.assertSameData(seqs, new_seqs)
batch_iter = sequtils.seq_batch_iter(seqs, 4)
self.assertIsInstance(batch_iter, types.GeneratorType)
batch_iter = list(batch_iter)
self.assertEqual(len(batch_iter), 2)
new_seqs = []
for seq_iter in batch_iter:
self.assertIsInstance(seq_iter, seqsift.utils.dataio.BufferedIter)
for s in seq_iter:
new_seqs.append(s)
self.assertSameData(seqs, new_seqs)
def test_odd(self):
seqs = [
SeqRecord(Seq('ATGACCAACTGA'), id='1'),
SeqRecord(Seq('ATGACCAACTGT'), id='2'),
SeqRecord(Seq('ATGACCAACTGC'), id='3'),
SeqRecord(Seq('ATGACCAACTGG'), id='4'),
SeqRecord(Seq('ATGACCAACTAT'), id='5'),
SeqRecord(Seq('ATGACCAACTAC'), id='6'),
SeqRecord(Seq('TTGACCAACTAC'), id='7'),
]
batch_iter = sequtils.seq_batch_iter(seqs, 3)
self.assertIsInstance(batch_iter, types.GeneratorType)
batch_iter = list(batch_iter)
self.assertEqual(len(batch_iter), 3)
new_seqs = []
for seq_iter in batch_iter:
self.assertIsInstance(seq_iter, seqsift.utils.dataio.BufferedIter)
for s in seq_iter:
new_seqs.append(s)
self.assertSameData(seqs, new_seqs)
batch_iter = sequtils.seq_batch_iter(seqs, 4)
self.assertIsInstance(batch_iter, types.GeneratorType)
batch_iter = list(batch_iter)
self.assertEqual(len(batch_iter), 2)
new_seqs = []
for seq_iter in batch_iter:
self.assertIsInstance(seq_iter, seqsift.utils.dataio.BufferedIter)
for s in seq_iter:
new_seqs.append(s)
self.assertSameData(seqs, new_seqs)
def get_seq_batch_iter_from_files(file_objs,
number_per_batch,
format = None,
data_type = 'dna',
ambiguities = True):
seqs = get_seq_iter(file_objs, format = format, data_type = data_type,
ambiguities = ambiguities)
return sequtils.seq_batch_iter(seqs, number_per_batch)
def get_seq_batch_iter_from_files(file_objs,
number_per_batch,
format=None,
data_type='dna',
ambiguities=True):
seqs = get_seq_iter(file_objs,
format=format,
data_type=data_type,
ambiguities=ambiguities)
return sequtils.seq_batch_iter(seqs, number_per_batch)