def setUp(self):
ref_seq = helpers.parse_fasta("""
>all
GGGGGGGGGTGTGTGTGT""")
self.aln = Alignment(helpers.parse_fasta("""
>seq1
GGGGGGGGGTGTGTGTGT
>seq2
AGAGAGAGGTGTGTGTGT
>seq3
GGGGGGGGGTATATATAT
""").values(), reference_sequence = ref_seq['all'])
def setUp(self):
ref_seq = helpers.parse_fasta("""
>all
GGGGGGGGGTGTGTGTGT""")
self.aln = Alignment(helpers.parse_fasta("""
>seq1
GGGGGGGGGTGTGTGTGT
>seq2
AGAGAGAGGTGTGTGTGT
>seq3
GGGGGGGGGTATATATAT
""").values(),
reference_sequence=ref_seq['all'])
def setUp(self):
ref_seq = Seq.Seq('GA' * 9 + 'GC' * 6 + 'GT' * 15)
query_seq = 'AA' * 4 + 'GA' * 5 + 'AC' + 'GC' * 5 + 'GT' * 15
aln_string = '>seq1\n{0}'.format(query_seq)
self.aln = Alignment(helpers.parse_fasta(aln_string).values(),
reference_sequence=ref_seq)
self.mutation_patterns = [mut_pattern.GA, mut_pattern.GM]
def main():
'''Main runner, to read data, compute and saves output.'''
sequences = parse_fasta(os.path.join(os.path.dirname(__file__), 'data/rosalind_grph.txt'))
adj_list = adjaceny_list(sequences, 3)
with open(output_path(__file__), 'w') as output:
output.write('\n'.join(adj_list))
def main(): '''Main runner, to read data, compute and saves output.''' sequences = parse_fasta(os.path.join(os.path.dirname(__file__), 'data/rosalind_gc.txt')) hi_gc = highest_gc(sequences) with open(output_path(__file__), 'w') as output: output.write(hi_gc['id'] + '\n') output.write(str(hi_gc['gc']))
def setUp(self):
aln_string = """
>seq1
GTCAGTCAGTCAGTCACCCC
>seq2
GTCAGTCAGTCAGTCACCCC
>seq3
ATCAATCAGTCAATCACCCC
"""
self.seqs = helpers.parse_fasta(aln_string)
self.aln = Alignment(self.seqs.values())
def setUp(self):
aln_string = """
>seq1
GGTGACGCT
>seq2
AGTAACGCT
>seq3
GGTAACACT
"""
ref_seq = Seq.Seq('GGTGACGCT')
self.aln = Alignment(helpers.parse_fasta(aln_string).values(), reference_sequence=ref_seq)
def setUp(self):
aln_string = """
>seq1
GTCAGTCAGTCAGTCACCCC
>seq2
GTCAGTCAGTCAGTCACCCC
>seq3
ATCAATCAGTCAATCACCCC
"""
self.seqs = helpers.parse_fasta(aln_string)
self.aln = Alignment(self.seqs.values())
def setUp(self):
aln_string = """
>seq1
GGTGACGCT
>seq2
AGTAACGCT
>seq3
GGTAACACT
"""
ref_seq = Seq.Seq('GGTGACGCT')
self.aln = Alignment(helpers.parse_fasta(aln_string).values(),
reference_sequence=ref_seq)
def main():
'''Main runner, to read data, compute and saves output.'''
sequences = parse_fasta(os.path.join(os.path.dirname(__file__), 'data/rosalind_cons.txt'))
matrix = profile_matrix(sequences)
consensus = concensus_string(matrix)
# add symbol labels to the profile matrix for printing
labeled_matrix = ['A:','C:','G:','T:']
for i, row in enumerate(matrix):
for col in row:
labeled_matrix[i] += ' ' + str(col)
with open(output_path(__file__), 'w') as output:
output.write(consensus + '\n')
output.write('\n'.join(labeled_matrix))
def test_with_reference_seqs(self):
ref_seqs = helpers.parse_fasta("""
>cluster1
AAAAAAAAAAAAAAAAACCC
>cluster2
CCTTGGCCGGTTGGCCGCCC
""")
aln_set = AlignmentSet(self.seqs, cluster_map=self.cluster_map,
reference_sequences=ref_seqs)
self.assertEqual(len(aln_set.clusters), 2)
for result in aln_set.multiple_context_analysis([old_pattern]):
hm_pos = result['call']['hm_pos']
if result['call']['sequence'] in ['seq2.{}'.format(i) for i in [1,2,3]]:
self.assertTrue(hm_pos)
else:
self.assertFalse(hm_pos)
def setUp(self):
aln_string = """
>seq1.1
GTCAGTCGGTCGGTCAGCCC
>seq1.2
GTCAGTCGGTCGGTCAGCCC
>seq1.3
AGCAATCAGGCAATCAGCCC
>seq2.1
GGTTAACCGGTTAACCGCCC
>seq2.2
GGTTAACCGGTTAACCGCCC
>seq2.3
AATTAACCAATTAACCACCC
"""
self.cluster_map = {'cluster1': ['seq1.1', 'seq1.2', 'seq1.3'],
'cluster2': ['seq2.1', 'seq2.2', 'seq2.3']}
self.seqs = helpers.parse_fasta(aln_string)
def test_with_reference_seqs(self):
ref_seqs = helpers.parse_fasta("""
>cluster1
AAAAAAAAAAAAAAAAACCC
>cluster2
CCTTGGCCGGTTGGCCGCCC
""")
aln_set = AlignmentSet(self.seqs,
cluster_map=self.cluster_map,
reference_sequences=ref_seqs)
self.assertEqual(len(aln_set.clusters), 2)
for result in aln_set.multiple_context_analysis([old_pattern]):
hm_pos = result['call']['hm_pos']
if result['call']['sequence'] in [
'seq2.{}'.format(i) for i in [1, 2, 3]
]:
self.assertTrue(hm_pos)
else:
self.assertFalse(hm_pos)
def setUp(self):
aln_string = """
>seq1.1
GTCAGTCGGTCGGTCAGCCC
>seq1.2
GTCAGTCGGTCGGTCAGCCC
>seq1.3
AGCAATCAGGCAATCAGCCC
>seq2.1
GGTTAACCGGTTAACCGCCC
>seq2.2
GGTTAACCGGTTAACCGCCC
>seq2.3
AATTAACCAATTAACCACCC
"""
self.cluster_map = {
'cluster1': ['seq1.1', 'seq1.2', 'seq1.3'],
'cluster2': ['seq2.1', 'seq2.2', 'seq2.3']
}
self.seqs = helpers.parse_fasta(aln_string)
def setUp(self):
ref_seq = Seq.Seq('GA'*9 + 'GC'*6 + 'GT'*15)
query_seq = 'AA'*4 + 'GA'*5 + 'AC' + 'GC'*5 + 'GT'*15
aln_string = '>seq1\n{0}'.format(query_seq)
self.aln = Alignment(helpers.parse_fasta(aln_string).values(), reference_sequence=ref_seq)
self.mutation_patterns=[mut_pattern.GA, mut_pattern.GM]
