def store_low_CAA_genes(genes):
""" Create list of genes where CAA usage < 0.9
"""
# compute codon usage
print('Computing codon statistics')
dnana = DNAAnalyzer(strict=False)
data = dnana.get_gene_codon_usages(genes)
def compute_norm(gene, *args):
""" Compute normalized occurrence frequency of aa
"""
all_codon_num = dnana._count_codons(str(gene.seq))
aa_num = sum([all_codon_num[codon] for codon in args])
norm = aa_num * 1000 / len(gene.seq)
return norm
avg_codon_freqs = dnana.get_codon_freqs(genes)
print(' LYS freq: %f\n' %
(avg_codon_freqs['AAA'] + avg_codon_freqs['AAG']) +
' GLU freq: %f\n' %
(avg_codon_freqs['GAA'] + avg_codon_freqs['GAG']) +
' GLN freq: %f' % (avg_codon_freqs['CAA'] + avg_codon_freqs['CAG']))
# filter for genes
low_CAA_genes = []
for gene, codu in data.items():
if not codu['CAA'] is None and codu['CAA'] < 0.9:
lys_freq = (compute_norm(gene, 'AAA', 'AAG') / 1000) / (
avg_codon_freqs['AAA'] + avg_codon_freqs['AAG'])
glu_freq = (compute_norm(gene, 'GAA', 'GAG') / 1000) / (
avg_codon_freqs['GAA'] + avg_codon_freqs['GAG'])
gln_freq = (compute_norm(gene, 'CAA', 'CAG') / 1000) / (
avg_codon_freqs['CAA'] + avg_codon_freqs['CAG'])
low_CAA_genes.append(
(gene.id, extract_gene_name(gene), lys_freq, codu['AAA'],
glu_freq, codu['GAA'], gln_freq, codu['CAA']))
# store results
with open('results/low_CAA_genes.csv', 'w') as fd:
wrtr = csv.writer(fd)
wrtr.writerow([
'ID', 'name', 'LYS rel freq', 'CU: AAA', 'GLU rel freq', 'CU: GAA',
'GLN rel freq', 'CU: CAA'
])
for entry in low_CAA_genes:
wrtr.writerow(entry)
class TestCodonUsage(TestCase):
def setUp(self):
self.dnana = DNAAnalyzer()
self.seq = 'AAAAAGAAA'
self.genes = [
SeqRecord(Seq('AAAAAAAAG')),
SeqRecord(Seq('AAAAAGAAA')),
SeqRecord(Seq('TTTTTCTTT')),
SeqRecord(Seq('TTCTTTTTC'))
]
def test_codon_counter(self):
count = self.dnana._count_codons(self.seq)
self.assertEqual(count['AAA'], 2)
self.assertEqual(count['AAG'], 1)
self.assertEqual(count['AAT'], 0)
self.assertEqual(count['AAC'], 0)
def test_codon_usage(self):
codu = self.dnana.get_codon_usage(self.seq)
self.assertEqual(round(codu['AAA'], 3), round(0.6666, 3))
self.assertEqual(round(codu['AAG'], 3), round(0.3333, 3))
self.assertEqual(codu['AAT'], None)
self.assertEqual(codu['AAC'], None)
def test_average_codon_usage(self):
avg_codu = self.dnana.get_avg_codon_usage(self.genes)
self.assertEqual(round(avg_codu['AAA'], 3), round(0.6666, 3))
self.assertEqual(round(avg_codu['AAG'], 3), round(0.3333, 3))
self.assertEqual(avg_codu['AAT'], None)
self.assertEqual(avg_codu['AAC'], None)
self.assertEqual(round(avg_codu['TTT'], 3), round(0.5, 3))
self.assertEqual(round(avg_codu['TTC'], 3), round(0.5, 3))
def test_codon_frequencies(self):
avg_cod_freqs = self.dnana.get_codon_freqs(self.genes)
self.assertEqual(round(avg_cod_freqs['AAA'], 3), 0.333)
self.assertEqual(round(avg_cod_freqs['AAG'], 3), 0.167)
self.assertEqual(round(avg_cod_freqs['TTT'], 3), 0.25)
self.assertEqual(round(avg_cod_freqs['TTC'], 3), 0.25)