def test_annotate_small_intronic_inversion(self):
gene = get_example_genes()['SVEP1']
reference_annotations = {gene.chr: [gene]}
reference_genome = {
gene.chr:
MockObject(seq=MockLongString(gene.seq, offset=gene.start - 1))
}
best = get_best(gene)
bpp = BreakpointPair(
Breakpoint('9', 113152627, 113152627, orient='L'),
Breakpoint('9', 113152635, 113152635, orient='L'),
opposing_strands=True,
stranded=False,
event_type=SVTYPE.INV,
protocol=PROTOCOL.GENOME,
untemplated_seq='',
)
annotations = annotate_events([bpp],
reference_genome=reference_genome,
annotations=reference_annotations)
for a in annotations:
print(a, a.transcript1, a.transcript2)
assert len(annotations) == 1
ann = annotations[0]
assert ann.transcript1 == best
assert ann.transcript2 == best
refseq = best.transcripts[0].get_seq(reference_genome)
assert len(ann.fusion.transcripts) == 1
assert ann.fusion.transcripts[0].get_seq() == refseq
def test_small_duplication(self):
bpp = BreakpointPair(
Breakpoint('6', 157100005, strand='+', orient='R'),
Breakpoint('6', 157100007, strand='+', orient='L'),
event_type=SVTYPE.DUP,
untemplated_seq='',
protocol=PROTOCOL.GENOME,
)
# annotate the breakpoint with the gene
annotations = annotate_events([bpp],
reference_genome=self.reference_genome,
annotations=self.reference_annotations)
self.assertEqual(1, len(annotations))
ann = Annotation(bpp, transcript1=self.best, transcript2=self.best)
ft = FusionTranscript.build(
ann,
self.reference_genome,
min_orf_size=300,
max_orf_cap=10,
min_domain_mapping_match=0.9,
)
ref_tx = self.best.translations[0]
fusion_tx = ft.translations[0]
# compare the fusion translation to the refernece translation to create the protein notation
ref_aa_seq = ref_tx.get_aa_seq(self.reference_genome)
call = IndelCall(ref_aa_seq, fusion_tx.get_aa_seq())
self.assertTrue(call.is_dup)
notation = call_protein_indel(ref_tx, fusion_tx, self.reference_genome)
print(notation)
self.assertEqual('ENST00000346085:p.G319dupG', notation)
def test_annotate_events_synonymous(self):
for gene_list in self.reference_annotations.values():
for gene in gene_list:
for t in gene.transcripts:
print(t)
b1 = Breakpoint(self.gene.chr,
95344068,
orient=ORIENT.LEFT,
strand=STRAND.NS)
b2 = Breakpoint(self.gene.chr,
95344379,
orient=ORIENT.RIGHT,
strand=STRAND.NS)
bpp = BreakpointPair(
b1,
b2,
stranded=False,
opposing_strands=False,
event_type=SVTYPE.DEL,
protocol=PROTOCOL.GENOME,
untemplated_seq='',
)
annotations = annotate_events([bpp],
reference_genome=self.reference_genome,
annotations=self.reference_annotations)
ann = annotations[0]
for a in annotations:
print(a, a.fusion, a.fusion.transcripts)
print(a.transcript1, a.transcript1.transcripts)
fseq = ann.fusion.transcripts[0].get_seq()
refseq = ann.transcript1.transcripts[0].get_seq(self.reference_genome)
self.assertEqual(refseq, fseq)
self.assertEqual(1, len(annotations))
def test_fusion_with_novel_splice_site(self):
bpp = BreakpointPair(
Breakpoint('7', 150268089, 150268089, 'L', '+'),
Breakpoint('8', 79715940, 79715940, 'L', '-'),
event_type=SVTYPE.ITRANS,
protocol=PROTOCOL.GENOME,
untemplated_seq='',
)
gimap4 = EXAMPLE_GENES['GIMAP4']
il7 = EXAMPLE_GENES['IL7']
ref_genome = {
gimap4.chr:
MockObject(seq=MockLongString(gimap4.seq, offset=gimap4.start -
1)),
il7.chr:
MockObject(seq=MockLongString(il7.seq, offset=il7.start - 1)),
}
annotations = annotate_events([bpp], {
gimap4.chr: [gimap4],
il7.chr: [il7]
}, ref_genome)
assert len(annotations) == 1
ann = annotations[0]
print(ann, ann.transcript1, ann.transcript2)
print(ann.fusion)
print(
ann.fusion.transcripts[0].splicing_pattern,
ann.fusion.transcripts[0].splicing_pattern.splice_type,
)
for ex in ann.fusion.transcripts[0].exons:
print(ex, len(ex))
assert False
def test_annotate_small_intronic_inversion(self):
bpp = BreakpointPair(
Breakpoint('9', 113152627, 113152627, orient='L'),
Breakpoint('9', 113152635, 113152635, orient='L'),
opposing_strands=True,
stranded=False,
event_type=SVTYPE.INV,
protocol=PROTOCOL.GENOME,
untemplated_seq=''
)
annotations = annotate_events([bpp], reference_genome=self.reference_genome, annotations=self.reference_annotations)
for a in annotations:
print(a, a.transcript1, a.transcript2)
self.assertEqual(1, len(annotations))
ann = annotations[0]
self.assertEqual(self.best, ann.transcript1)
self.assertEqual(self.best, ann.transcript2)
refseq = self.best.transcripts[0].get_seq(self.reference_genome)
self.assertEqual(1, len(ann.fusion.transcripts))
self.assertEqual(refseq, ann.fusion.transcripts[0].get_seq())
def test_small_duplication(self):
gene = get_example_genes()['ARID1B']
reference_annotations = {gene.chr: [gene]}
reference_genome = {
gene.chr:
MockObject(seq=MockLongString(gene.seq, offset=gene.start - 1))
}
best = get_best(gene)
bpp = BreakpointPair(
Breakpoint('6', 157100005, strand='+', orient='R'),
Breakpoint('6', 157100007, strand='+', orient='L'),
event_type=SVTYPE.DUP,
untemplated_seq='',
protocol=PROTOCOL.GENOME,
)
# annotate the breakpoint with the gene
annotations = annotate_events([bpp],
reference_genome=reference_genome,
annotations=reference_annotations)
assert len(annotations) == 1
ann = Annotation(bpp, transcript1=best, transcript2=best)
ft = FusionTranscript.build(
ann,
reference_genome,
min_orf_size=300,
max_orf_cap=10,
min_domain_mapping_match=0.9,
)
ref_tx = best.translations[0]
fusion_tx = ft.translations[0]
# compare the fusion translation to the refernece translation to create the protein notation
ref_aa_seq = ref_tx.get_aa_seq(reference_genome)
call = IndelCall(ref_aa_seq, fusion_tx.get_aa_seq())
assert call.is_dup
notation = call_protein_indel(ref_tx, fusion_tx, reference_genome)
print(notation)
assert notation == 'ENST00000346085:p.G319dupG'
def test_annotate_events_synonymous(self):
gene = get_example_genes()['NDUFA12']
reference_annotations = {gene.chr: [gene]}
reference_genome = {
gene.chr:
MockObject(seq=MockLongString(gene.seq, offset=gene.start - 1))
}
for gene_list in reference_annotations.values():
for gene in gene_list:
for t in gene.transcripts:
print(t)
b1 = Breakpoint(gene.chr,
95344068,
orient=ORIENT.LEFT,
strand=STRAND.NS)
b2 = Breakpoint(gene.chr,
95344379,
orient=ORIENT.RIGHT,
strand=STRAND.NS)
bpp = BreakpointPair(
b1,
b2,
stranded=False,
opposing_strands=False,
event_type=SVTYPE.DEL,
protocol=PROTOCOL.GENOME,
untemplated_seq='',
)
annotations = annotate_events([bpp],
reference_genome=reference_genome,
annotations=reference_annotations)
ann = annotations[0]
for a in annotations:
print(a, a.fusion, a.fusion.transcripts)
print(a.transcript1, a.transcript1.transcripts)
fseq = ann.fusion.transcripts[0].get_seq()
refseq = ann.transcript1.transcripts[0].get_seq(reference_genome)
assert fseq == refseq
assert len(annotations) == 1
def test_fusion_with_novel_splice_site(self):
raise unittest.SkipTest('TODO: dependent functionality not yet implemented')
bpp = BreakpointPair(
Breakpoint('7', 150268089, 150268089, 'L', '+'),
Breakpoint('8', 79715940, 79715940, 'L', '-'),
event_type=SVTYPE.ITRANS,
protocol=PROTOCOL.GENOME,
untemplated_seq=''
)
gimap4 = EXAMPLE_GENES['GIMAP4']
il7 = EXAMPLE_GENES['IL7']
ref_genome = {
gimap4.chr: MockObject(seq=MockLongString(gimap4.seq, offset=gimap4.start - 1)),
il7.chr: MockObject(seq=MockLongString(il7.seq, offset=il7.start - 1))
}
annotations = annotate_events([bpp], {gimap4.chr: [gimap4], il7.chr: [il7]}, ref_genome)
self.assertEqual(1, len(annotations))
ann = annotations[0]
print(ann, ann.transcript1, ann.transcript2)
print(ann.fusion)
print(ann.fusion.transcripts[0].splicing_pattern, ann.fusion.transcripts[0].splicing_pattern.splice_type)
for ex in ann.fusion.transcripts[0].exons:
print(ex, len(ex))
self.assertTrue(False)
