def test_mpileup_test2(self):
test = open(ospj(DATA_PATH, "mpileup_test2.tsv")).read()
bpdf = sufam.__main__.get_baseparser_extended_df("test", [mpileup_parser.parse(test)], "G", "GAA")
assert_equals(int(bpdf['cov'].iloc[0]), int(bpdf.G.iloc[0]))
assert_equals(test.count(",") + test.count("."), int(bpdf['cov'].iloc[0]))
assert_almost_equals(0.4324, float(bpdf.val_maf.iloc[0]), places=3)
assert_almost_equals(0.4324, float(bpdf.most_common_indel_maf.iloc[0]), places=3)
assert_equals("+", bpdf.most_common_indel_type.iloc[0])
def test_mpileup_test3(self):
test = open(ospj(DATA_PATH, "mpileup_test3.tsv")).read()
bpdf = sufam.__main__.get_baseparser_extended_df("test", [mpileup_parser.parse(test)], "G", "A")
assert_equals(int(bpdf['cov'].iloc[0]), int(bpdf.G.iloc[0]) + int(bpdf.A.iloc[0]) + int(bpdf["T"].iloc[0]))
assert_equals(1, int(bpdf["T"].iloc[0]))
assert_equals("AA", bpdf.most_common_indel.iloc[0])
assert_equals("+", bpdf.most_common_indel_type.iloc[0])
assert_almost_equals(0.0139, float(bpdf.val_maf.iloc[0]), places=3)
assert_almost_equals(0.0139, float(bpdf.most_common_al_maf.iloc[0]), places=3)
def test_mpileup_parser_get_mutations(self):
indel_followed_by_snv = "X\t150349557\tC\t28\t.$,-12caccactggccat-12CACCACTGGCCAT.,.,,,,.-12CACCACTGGCCAT,..,,-12caccactggccaG..-12CACCACTGGCCAA,,,..,\t;FCDDDDDDD/FDCC/C/E<FBDC\n"
muts = mpileup_parser.get_mutations(indel_followed_by_snv)
assert_equals(len(muts), 4)
assert_equals(muts.snvs["A"].count, 1)
assert_equals(muts.snvs["G"].count, 1)
assert_equals(muts.snvs["T"].count, 3)
assert_equals(muts.deletions["CACCACTGGCCA"].count, 5)
assert_almost_equals(muts.deletions["CACCACTGGCCA"].maf, 5/28.0, places=3)
assert_equals("X\t150349557\tC\t28\t1\t23\t1\t3\t0\tCACCACTGGCCA,CACCACTGGCCA,CACCACTGGCCA,CACCACTGGCCA,CACCACTGGCCA\t",
mpileup_parser.parse(indel_followed_by_snv))
def test_mpileup_test2(self):
test = open(ospj(DATA_PATH, "mpileup_test2.tsv")).read()
bpdf = sufam.__main__.get_baseparser_extended_df(
"test", [mpileup_parser.parse(test)], "G", "GAA")
assert_equals(int(bpdf['cov'].iloc[0]), int(bpdf.G.iloc[0]))
assert_equals(
test.count(",") + test.count("."), int(bpdf['cov'].iloc[0]))
assert_almost_equals(0.4324, float(bpdf.val_maf.iloc[0]), places=3)
assert_almost_equals(0.4324,
float(bpdf.most_common_indel_maf.iloc[0]),
places=3)
assert_equals("+", bpdf.most_common_indel_type.iloc[0])
def test_mpileup_parser_get_mutations(self):
indel_followed_by_snv = "X\t150349557\tC\t28\t.$,-12caccactggccat-12CACCACTGGCCAT.,.,,,,.-12CACCACTGGCCAT,..,,-12caccactggccaG..-12CACCACTGGCCAA,,,..,\t;FCDDDDDDD/FDCC/C/E<FBDC\n"
muts = mpileup_parser.get_mutations(indel_followed_by_snv)
assert_equals(len(muts), 4)
assert_equals(muts.snvs["A"].count, 1)
assert_equals(muts.snvs["G"].count, 1)
assert_equals(muts.snvs["T"].count, 3)
assert_equals(muts.deletions["CACCACTGGCCA"].count, 5)
assert_almost_equals(muts.deletions["CACCACTGGCCA"].maf,
5 / 28.0,
places=3)
assert_equals(
"X\t150349557\tC\t28\t1\t23\t1\t3\t0\tCACCACTGGCCA,CACCACTGGCCA,CACCACTGGCCA,CACCACTGGCCA,CACCACTGGCCA\t",
mpileup_parser.parse(indel_followed_by_snv))
def test_mpileup_test3(self):
test = open(ospj(DATA_PATH, "mpileup_test3.tsv")).read()
bpdf = sufam.__main__.get_baseparser_extended_df(
"test", [mpileup_parser.parse(test)], "G", "A")
assert_equals(
int(bpdf['cov'].iloc[0]),
int(bpdf.G.iloc[0]) + int(bpdf.A.iloc[0]) + int(bpdf["T"].iloc[0]))
assert_equals(1, int(bpdf["T"].iloc[0]))
assert_equals("AA", bpdf.most_common_indel.iloc[0])
assert_equals("+", bpdf.most_common_indel_type.iloc[0])
assert_almost_equals(0.0139, float(bpdf.val_maf.iloc[0]), places=3)
assert_almost_equals(0.0139,
float(bpdf.most_common_al_maf.iloc[0]),
places=3)
def test_mpileup_parser_indel_followed_by_snv(self):
indel_followed_by_snv = "X\t150349557\tC\t28\t.$,-12caccactggccat-12CACCACTGGCCAT.,.,,,,.-12CACCACTGGCCAT,..,,-12caccactggccaG..-12CACCACTGGCCAA,,,..,\t;FCDDDDDDD/FDCC/C/E<FBDC\n"
assert_equals(
"X\t150349557\tC\t28\t1\t23\t1\t3\t0\tCACCACTGGCCA,CACCACTGGCCA,CACCACTGGCCA,CACCACTGGCCA,CACCACTGGCCA\t",
mpileup_parser.parse(indel_followed_by_snv))
def test_mpileup_parser_two_digit_indel(self):
two_digit_indel = "X\t150349557\tC\t24\t.$,-12caccactggcca.-12CACCACTGGCCA.,.,,,,.-12CACCACTGGCCA,..,,-12caccactggcca..-12CACCACTGGCCA,,,..,\t;FCDDDDDDD/FDCC/C/E<FBDC\n"
assert_equals(
"X\t150349557\tC\t24\t0\t24\t0\t0\t0\tCACCACTGGCCA,CACCACTGGCCA,CACCACTGGCCA,CACCACTGGCCA,CACCACTGGCCA\t",
mpileup_parser.parse(two_digit_indel))
def test_0cov_regionwithcov(self):
test = open(ospj(DATA_PATH, "mpileup_0cov_regionwithcov.tsv")).read()
bpdf = sufam.__main__.get_baseparser_extended_df(
"test", [mpileup_parser.parse(test)], "G", "GAA")
assert_equals(None, bpdf)
def test_mpileup_parser_indel_followed_by_snv(self):
indel_followed_by_snv = "X\t150349557\tC\t28\t.$,-12caccactggccat-12CACCACTGGCCAT.,.,,,,.-12CACCACTGGCCAT,..,,-12caccactggccaG..-12CACCACTGGCCAA,,,..,\t;FCDDDDDDD/FDCC/C/E<FBDC\n"
assert_equals("X\t150349557\tC\t28\t1\t23\t1\t3\t0\tCACCACTGGCCA,CACCACTGGCCA,CACCACTGGCCA,CACCACTGGCCA,CACCACTGGCCA\t",
mpileup_parser.parse(indel_followed_by_snv))
def test_mpileup_parser_two_digit_indel(self):
two_digit_indel = "X\t150349557\tC\t24\t.$,-12caccactggcca.-12CACCACTGGCCA.,.,,,,.-12CACCACTGGCCA,..,,-12caccactggcca..-12CACCACTGGCCA,,,..,\t;FCDDDDDDD/FDCC/C/E<FBDC\n"
assert_equals("X\t150349557\tC\t24\t0\t24\t0\t0\t0\tCACCACTGGCCA,CACCACTGGCCA,CACCACTGGCCA,CACCACTGGCCA,CACCACTGGCCA\t",
mpileup_parser.parse(two_digit_indel))
def test_0cov_regionwithcov(self):
test = open(ospj(DATA_PATH, "mpileup_0cov_regionwithcov.tsv")).read()
bpdf = sufam.__main__.get_baseparser_extended_df("test", [mpileup_parser.parse(test)], "G", "GAA")
assert_equals(None, bpdf)