def test_g2x_noncoding_reverse():
"""
Do some g. to n. conversion checking for the gene on the reverse
strand.
"""
rna = [2000, 2797, 6405, 6528, 31678, 31750, 47687, 47806, 76448,
76535, 92453, 92554, 123276, 123470, 146090, 146213]
cm = Crossmap(rna, [], -1)
# Fix for r536: disable the -u and +d convention.
# assert cm.tuple2string(cm.g2x(146214)) == '1-u1'
assert cm.tuple2string(cm.g2x(146214)) == '-1'
assert cm.tuple2string(cm.g2x(146213)) == '1'
assert cm.tuple2string(cm.g2x(146089)) == '124+1'
# Fix for r536: disable the -u and +d convention.
# assert cm.tuple2string(cm.g2x(1999)) == '1624+d1'
assert cm.tuple2string(cm.g2x(1999)) == '*1'
def test_g2x_noncoding():
"""
Do some g. to n. conversion checking for the gene on the forward
strand.
"""
rna = [5002, 5125, 27745, 27939, 58661, 58762, 74680, 74767, 103409,
103528, 119465, 119537, 144687, 144810, 148418, 149215]
cm = Crossmap(rna, [], 1)
# Fix for r536: disable the -u and +d convention.
# assert cm.tuple2string(cm.g2x(5001)) == '1-u1'
assert cm.tuple2string(cm.g2x(5001)) == '-1'
assert cm.tuple2string(cm.g2x(5002)) == '1'
assert cm.tuple2string(cm.g2x(5126)) == '124+1'
# Fix for r536: disable the -u and +d convention.
# assert cm.tuple2string(cm.g2x(149216)) == '1624+d1'
assert cm.tuple2string(cm.g2x(149216)) == '*1'
def test_cds_one_exon():
"""
Test a gene that has a CDS that lies entirely in one exon.
"""
rna = [1, 80, 81, 3719]
cds = [162, 2123]
cm = Crossmap(rna, cds, 1)
assert cm._Crossmap__crossmapping == [-161, -82, -81, 3558]
assert cm.x2g(1, 0) == 162
assert cm.tuple2string(cm.g2x(2123)) == '1962'
assert cm.tuple2string(cm.g2x(2124)) == '*1'
def test_cds_start_on_splice_site():
"""
Test a gene that has a CDS that starts on an exon splice site.
"""
rna = [23755059, 23755214, 23777833, 23778028, 23808749, 23808851,
23824768, 23824856, 23853497, 23853617, 23869553, 23869626,
23894775, 23894899, 23898506, 23899304]
cds = [23777833, 23898680]
cm = Crossmap(rna, cds, 1)
assert (cm._Crossmap__crossmapping ==
[-156, -1, 1, 196, 197, 299, 300, 388, 389, 509, 510, 583,
584, 708, 709, 1507])
assert cm.x2g(1, 0) == 23777833
# Fix for r536: disable the -u and +d convention.
# assert cm.tuple2string(cm.g2x(2123)) == '-156-u23752936'
# assert cm.tuple2string(cm.g2x(2124)) == '-156-u23752935'
assert cm.tuple2string(cm.g2x(2123)) == '-23753092'
assert cm.tuple2string(cm.g2x(2124)) == '-23753091'
def test_g2x_reverse():
"""
Do some g. to c. conversion checking for the gene on the reverse
strand.
"""
rna = [2000, 2797, 6405, 6528, 31678, 31750, 47687, 47806, 76448,
76535, 92453, 92554, 123276, 123470, 146090, 146213]
cds = [76479, 123290]
cm = Crossmap(rna, cds, -1)
# Fix for r536: disable the -u and +d convention.
# assert cm.tuple2string(cm.g2x(146214)) == '-304-u1'
assert cm.tuple2string(cm.g2x(146214)) == '-305'
assert cm.tuple2string(cm.g2x(146091)) == '-182'
assert cm.tuple2string(cm.g2x(146089)) == '-181+1'
assert cm.tuple2string(cm.g2x(123291)) == '-1'
assert cm.tuple2string(cm.g2x(123290)) == '1'
assert cm.tuple2string(cm.g2x(92555)) == '16-1'
assert cm.tuple2string(cm.g2x(76479)) == '174'
assert cm.tuple2string(cm.g2x(76478)) == '*1'
assert cm.tuple2string(cm.g2x(47807)) == '*32-1'
assert cm.tuple2string(cm.g2x(47805)) == '*33'
# Fix for r536: disable the -u and +d convention.
# assert cm.tuple2string(cm.g2x(1999)) == '*1146+d1'
assert cm.tuple2string(cm.g2x(1999)) == '*1147'
def test_g2x():
"""
Do some g. to c. conversion checking for the gene on the forward
strand.
"""
rna = [5002, 5125, 27745, 27939, 58661, 58762, 74680, 74767, 103409,
103528, 119465, 119537, 144687, 144810, 148418, 149215]
cds = [27925, 74736]
cm = Crossmap(rna, cds, 1)
# Fix for r536: disable the -u and +d convention.
# assert cm.tuple2string(cm.g2x(5001)) == '-304-u1'
assert cm.tuple2string(cm.g2x(5001)) == '-305'
assert cm.tuple2string(cm.g2x(5124)) == '-182'
assert cm.tuple2string(cm.g2x(5126)) == '-181+1'
assert cm.tuple2string(cm.g2x(27924)) == '-1'
assert cm.tuple2string(cm.g2x(27925)) == '1'
assert cm.tuple2string(cm.g2x(58660)) == '16-1'
assert cm.tuple2string(cm.g2x(74736)) == '174'
assert cm.tuple2string(cm.g2x(74737)) == '*1'
assert cm.tuple2string(cm.g2x(103408)) == '*32-1'
assert cm.tuple2string(cm.g2x(103410)) == '*33'
# Fix for r536: disable the -u and +d convention.
# assert cm.tuple2string(cm.g2x(149216)) == '*1146+d1'
assert cm.tuple2string(cm.g2x(149216)) == '*1147'
