def test_get_uorf_start_stop_baisc_negative(self):
"""
Test 3 reading frames orf detection, negative strand
"""
intervals = pybedtools.BedTool(""" chr1 1 9 ENSG1 0 -
chr1 1 10 ENSG2 0 -
chr1 1 11 ENSG3 0 -
""", from_string=True)
test_dict = {
"ENSG1" : [(intervals[0], SeqRecord(Seq("ATGGGGTAG", IUPAC.unambiguous_dna)))],
"ENSG2" : [(intervals[1], SeqRecord(Seq("AATGGGGTAG", IUPAC.unambiguous_dna)))],
"ENSG3" : [(intervals[2], SeqRecord(Seq("AAATGGGGTAG", IUPAC.unambiguous_dna)))],
}
detector = UORF_detector()
test_result = detector._get_uorf_start_stop(test_dict, uorf_length=0)
true_result = pybedtools.BedTool(
[
pybedtools.create_interval_from_list(["chr1", "protein_coding",
"uORF_start", "7", "9", ".",
"-", ".",
"ID=uORF_start:ENSG1;Parent=ENSG1" ]),
pybedtools.create_interval_from_list(["chr1", "protein_coding",
"uORF_end", "1", "3", ".",
"-", ".",
"ID=uORF_end:ENSG1;Parent=ENSG1" ]),
pybedtools.create_interval_from_list(["chr1", "protein_coding",
"uORF_start", "7", "9", ".",
"-", ".",
"ID=uORF_start:ENSG2;Parent=ENSG2" ]),
pybedtools.create_interval_from_list(["chr1", "protein_coding",
"uORF_end", "1", "3", ".",
"-", ".",
"ID=uORF_end:ENSG2;Parent=ENSG2" ]),
pybedtools.create_interval_from_list(["chr1", "protein_coding",
"uORF_start", "7", "9", ".",
"-", ".",
"ID=uORF_start:ENSG3;Parent=ENSG3" ]),
pybedtools.create_interval_from_list(["chr1", "protein_coding",
"uORF_end", "1", "3", ".",
"-", ".",
"ID=uORF_end:ENSG3;Parent=ENSG3" ]),
]
)
test_result = [str(x) for x in test_result]
true_result = [str(x) for x in true_result]
self.assertListEqual(test_result, true_result)
def test_get_uorf_start_stop_split_negative(self):
"""
start / stop codon detection spanning introns negative strand
"""
intervals = pybedtools.BedTool(""" chr1 12 18 ENSG1 0 -
chr1 1 6 ENSG1 0 -
chr1 12 18 ENSG2 0 -
chr1 1 6 ENSG2 0 -
chr1 12 18 ENSG3 0 -
chr1 1 6 ENSG3 0 -
""", from_string=True)
test_dict = {
"ENSG1" : [(intervals[0], SeqRecord(Seq("GGGATG", IUPAC.unambiguous_dna))),
(intervals[1], SeqRecord(Seq("GGGTAG", IUPAC.unambiguous_dna)))],
"ENSG2" : [(intervals[2], SeqRecord(Seq("GATGGG", IUPAC.unambiguous_dna))),
(intervals[3], SeqRecord(Seq("GTAGGG", IUPAC.unambiguous_dna)))],
"ENSG3" : [(intervals[4], SeqRecord(Seq("GGATGG", IUPAC.unambiguous_dna))),
(intervals[5], SeqRecord(Seq("GGTAGG", IUPAC.unambiguous_dna)))]
}
detector = UORF_detector()
test_result = detector._get_uorf_start_stop(test_dict, uorf_length=0)
true_result = pybedtools.BedTool(
[
pybedtools.create_interval_from_list(["chr1", "protein_coding",
"uORF_start", "13", "15", ".",
"-", ".",
"ID=uORF_start:ENSG1;Parent=ENSG1" ]),
pybedtools.create_interval_from_list(["chr1", "protein_coding",
"uORF_end", "1", "3", ".",
"-", ".",
"ID=uORF_end:ENSG1;Parent=ENSG1" ]),
pybedtools.create_interval_from_list(["chr1", "protein_coding",
"uORF_start", "15", "17", ".",
"-", ".",
"ID=uORF_start:ENSG2;Parent=ENSG2" ]),
pybedtools.create_interval_from_list(["chr1", "protein_coding",
"uORF_end", "3", "5", ".",
"-", ".",
"ID=uORF_end:ENSG2;Parent=ENSG2" ]),
pybedtools.create_interval_from_list(["chr1", "protein_coding",
"uORF_start", "14", "16", ".",
"-", ".",
"ID=uORF_start:ENSG3;Parent=ENSG3" ]),
pybedtools.create_interval_from_list(["chr1", "protein_coding",
"uORF_end", "2", "4", ".",
"-", ".",
"ID=uORF_end:ENSG3;Parent=ENSG3" ]),
]
)
true_result = [str(interval) for interval in true_result]
test_result = [str(interval) for interval in test_result]
print test_result
print true_result
self.assertEqual(test_result, true_result)
