def test_intron_annotation(self):
"""sequences annotated with introns should return correct seq"""
for symbol, stable_id, rank, exp_seq5, exp_seq3 in [
("IL2", "ENST00000226730", 1, "gtaagtatat", "actttcttag"),
("IL13", "ENST00000304506", 3, "gtaaggcatc", "tgtcctgcag"),
]:
gene = asserted_one(self.human.get_genes_matching(symbol=symbol))
seq = gene.get_annotated_seq(feature_types="gene")
intron = asserted_one(
seq.get_annotations_matching("intron",
"%s-%d" % (stable_id, rank)))
intron_seq = str(seq.get_region_covering_all(intron).get_slice())
self.assertEqual(intron_seq[:10], exp_seq5.upper())
self.assertEqual(intron_seq[-10:], exp_seq3.upper())
def test_intron_number(self):
"""number of introns should be correct"""
for gene_id, transcript_id, exp_number in [
("ENSG00000227268", "ENST00000445946", 0),
("ENSG00000132199", "ENST00000583771", 5),
("ENSG00000132199", "ENST00000340116", 14),
]:
gene = asserted_one(
self.human.get_genes_matching(stableid=gene_id))
transcript = asserted_one(
[t for t in gene.transcripts if t.stableid == transcript_id])
if exp_number == 0:
self.assertEqual(transcript.introns, None)
else:
self.assertEqual(len(transcript.introns), exp_number)
def test_intron(self):
"""should get correct Intron sequence, regardless of strand"""
# IL2 is on - strand, IL13 is on + strand, both have three introns
IL2_exp_introns = [
(1, 122456203, 122456293, "gtaagtatat", "actttcttag"),
(2, 122453853, 122456143, "gtaagtacaa", "attattctag"),
(3, 122451862, 122453709, "gtaaggcatt", "tcttttatag"),
]
IL13_exp_introns = [
(1, 132658360, 132659417, "gtgagtgtcg", "gctcccacag"),
(2, 132659471, 132659723, "gtaaggacct", "ctccccacag"),
(3, 132659828, 132660174, "gtaaggcatc", "tgtcctgcag"),
]
for symbol, stable_id, exp_introns in [
("IL2", "ENST00000226730", IL2_exp_introns),
("IL13", "ENST00000304506", IL13_exp_introns),
]:
gene = asserted_one(self.human.get_genes_matching(symbol=symbol))
strand = gene.location.strand
transcript = asserted_one(
[t for t in gene.transcripts if t.stableid == stable_id])
introns = transcript.introns
self.assertEqual(len(introns), len(exp_introns))
idx = 0
for intron in introns:
loc = intron.location
start, end = loc.start, loc.end
seq = str(intron.seq)
exp_rank, exp_start, exp_end, exp_seq5, exp_seq3 = exp_introns[
idx]
self.assertEqual(loc.strand, strand)
# test the order using rank
self.assertEqual(intron.rank, exp_rank)
# test position
self.assertEqual(start, exp_start)
self.assertEqual(end, exp_end)
# test sequence
self.assertEqual(seq[:10], exp_seq5.upper())
self.assertEqual(seq[-10:], exp_seq3.upper())
idx += 1