def test_get_de_novo_counts_loss_of_function(self):
""" check the de novo counts for loss-of-function consequences
"""
# check all the loss-of-function-equivalent SNV VEP consequences
snv = [
"stop_gained", "splice_acceptor_variant", "splice_donor_variant",
"initiator_codon_variant", "start_lost",
"conserved_exon_terminus_variant"
]
self.expected["lof_snv"] = 1
for cq in snv:
self.de_novos["consequence"] = cq
computed = get_de_novo_counts(self.de_novos)
self.compare_tables(computed, self.expected)
# check all the loss-of-function-equivalent inframe insertion and
# deletion VEP consequences
indel = ["frameshift_variant"]
self.expected["lof_snv"] = 0
self.expected["lof_indel"] = 1
self.de_novos["ref"] = "GG"
for cq in indel:
self.de_novos["consequence"] = cq
computed = get_de_novo_counts(self.de_novos)
self.compare_tables(computed, self.expected)
def test_get_de_novo_counts_loss_of_function(self):
""" check the de novo counts for loss-of-function consequences
"""
# check all the loss-of-function-equivalent SNV VEP consequences
snv = ["stop_gained", "splice_acceptor_variant", "splice_donor_variant",
"initiator_codon_variant", "start_lost",
"conserved_exon_terminus_variant"]
self.expected["lof_snv"] = 1
for cq in snv:
self.de_novos["consequence"] = cq
computed = get_de_novo_counts(self.de_novos)
self.compare_tables(computed, self.expected)
# check all the loss-of-function-equivalent inframe insertion and
# deletion VEP consequences
indel = ["frameshift_variant"]
self.expected["lof_snv"] = 0
self.expected["lof_indel"] = 1
self.de_novos["ref"] = "GG"
for cq in indel:
self.de_novos["consequence"] = cq
computed = get_de_novo_counts(self.de_novos)
self.compare_tables(computed, self.expected)
def test_get_de_novo_counts_missense(self):
""" check the de novo counts for missense-equivalent consequences
"""
# check all the possible missense-equivalent SNV VEP consequences
snv = [
"missense_variant", "stop_lost", "coding_sequence_variant",
"protein_altering_variant"
]
self.expected["missense_snv"] = 1
for cq in snv:
self.de_novos["consequence"] = cq
computed = get_de_novo_counts(self.de_novos)
self.compare_tables(computed, self.expected)
# check all the possible missense-equivalent inframe insertion and
# deletion VEP consequences
indel = ["inframe_deletion", "inframe_insertion"]
self.expected["missense_snv"] = 0
self.expected["missense_indel"] = 1
self.de_novos["ref"] = "GG"
for cq in indel:
self.de_novos["consequence"] = cq
computed = get_de_novo_counts(self.de_novos)
self.compare_tables(computed, self.expected)
def test_get_de_novo_counts_multiple(self):
""" check the de novo counts for multiple variants
"""
# add another variant to the dataframe
new_row = self.de_novos.copy()
self.de_novos = self.de_novos.append(new_row, ignore_index=True)
# counting two of the same variant give a count of 2
self.expected["missense_snv"] = 2
self.de_novos["consequence"] = "missense_variant"
computed = get_de_novo_counts(self.de_novos)
self.compare_tables(computed, self.expected)
# counting two different variant types gives two counts of 1
self.expected["missense_snv"] = 1
self.expected["lof_snv"] = 1
self.de_novos.loc[0, "consequence"] = "stop_gained"
computed = get_de_novo_counts(self.de_novos)
self.compare_tables(computed, self.expected)
# now check when we have two variants in different genes
new_row = self.expected.copy()
self.expected = self.expected.append(new_row, ignore_index=True)
self.de_novos.loc[0, "hgnc"] = "ARID1B"
self.expected.loc[0, "hgnc"] = "ARID1B"
self.expected.loc[0, "missense_snv"] = 0
self.expected.loc[1, "lof_snv"] = 0
computed = get_de_novo_counts(self.de_novos)
self.compare_tables(computed, self.expected)
def test_get_de_novo_counts_multiple(self):
""" check the de novo counts for multiple variants
"""
# add another variant to the dataframe
new_row = self.de_novos.copy()
self.de_novos = self.de_novos.append(new_row, ignore_index=True)
# counting two of the same variant give a count of 2
self.expected["missense_snv"] = 2
self.de_novos["consequence"] = "missense_variant"
computed = get_de_novo_counts(self.de_novos)
self.compare_tables(computed, self.expected)
# counting two different variant types gives two counts of 1
self.expected["missense_snv"] = 1
self.expected["lof_snv"] = 1
self.de_novos.loc[0, "consequence"] = "stop_gained"
computed = get_de_novo_counts(self.de_novos)
self.compare_tables(computed, self.expected)
# now check when we have two variants in different genes
new_row = self.expected.copy()
self.expected = self.expected.append(new_row, ignore_index=True)
self.de_novos.loc[0, "hgnc"] = "ARID1B"
self.expected.loc[0, "hgnc"] = "ARID1B"
self.expected.loc[0, "missense_snv"] = 0
self.expected.loc[1, "lof_snv"] = 0
computed = get_de_novo_counts(self.de_novos)
self.compare_tables(computed, self.expected)
def test_get_de_novo_counts_missense(self):
""" check the de novo counts for missense-equivalent consequences
"""
# check all the possible missense-equivalent SNV VEP consequences
snv = ["missense_variant", "stop_lost", "coding_sequence_variant",
"protein_altering_variant"]
self.expected["missense_snv"] = 1
for cq in snv:
self.de_novos["consequence"] = cq
computed = get_de_novo_counts(self.de_novos)
self.compare_tables(computed, self.expected)
# check all the possible missense-equivalent inframe insertion and
# deletion VEP consequences
indel = ["inframe_deletion", "inframe_insertion"]
self.expected["missense_snv"] = 0
self.expected["missense_indel"] = 1
self.de_novos["ref"] = "GG"
for cq in indel:
self.de_novos["consequence"] = cq
computed = get_de_novo_counts(self.de_novos)
self.compare_tables(computed, self.expected)
def check_enrichment(constraint, de_novos, cache_dir, male, female, threshold,
ratio):
rates = get_rates_by_constraint(constraint, cache_dir, threshold, ratio)
constrained_exp = get_expected_mutations(
prepare_rates(rates['constrained']), male, female)
unconstrained_exp = get_expected_mutations(
prepare_rates(rates['unconstrained']), male, female)
in_constraint = classify_de_novos_by_constraint(constraint, de_novos,
cache_dir, threshold,
ratio)
constrained_obs = get_de_novo_counts(de_novos[in_constraint])
unconstrained_obs = get_de_novo_counts(
de_novos[[not x for x in in_constraint]])
constrained_enrich = enrichment(constrained_obs, constrained_exp)
unconstrained_enrich = enrichment(unconstrained_obs, unconstrained_exp)
ptv_diff = compare_regions(constrained_enrich, unconstrained_enrich, 'PTV')
pav_diff = compare_regions(constrained_enrich, unconstrained_enrich, 'PAV')
return {
'constrained_enrich': constrained_enrich,
'unconstrained_enrich': unconstrained_enrich,
'PTV diff': ptv_diff,
'PAV diff': pav_diff
}
def analyse_enrichment(de_novos, trios, rates=None, plot_path=None):
""" analyse whether de novo mutations are enriched in genes
Args:
de_novos: data frame containing all the observed de novos for all the
genes
trios: dictionary of male and female proband counts in the population
plot_path: path to save enrichment plots to, or None
rates: gene-based mutation rates data frame, or None
Returns:
data frame containing results from testing for enrichment of de
in each gene with de novos in it.
"""
observed = get_de_novo_counts(de_novos)
expected = get_expected_mutations(rates, trios["male"], trios["female"])
# calculate p values for each gene using the mutation rates
enrichment = gene_enrichment(expected, observed)
# make a manhattan plot of enrichment P values
if plot_path is not None:
num_tests = 18500
plot_enrichment(enrichment,
num_tests,
plot_path,
p_columns=["p_lof", "p_func"])
# remove the position column (which is only used to be able to locate the
# gene's position on a chromosome on a Manhattan plot).
del enrichment["start_pos"]
return enrichment
def analyse_enrichment(de_novos, trios, rates=None, plot_path=None):
""" analyse whether de novo mutations are enriched in genes
Args:
de_novos: data frame containing all the observed de novos for all the
genes
trios: dictionary of male and female proband counts in the population
plot_path: path to save enrichment plots to, or None
rates: gene-based mutation rates data frame, or None
Returns:
data frame containing results from testing for enrichment of de
in each gene with de novos in it.
"""
observed = get_de_novo_counts(de_novos)
expected = get_expected_mutations(rates, trios["male"], trios["female"])
# calculate p values for each gene using the mutation rates
enrichment = gene_enrichment(expected, observed)
# make a manhattan plot of enrichment P values
if plot_path is not None:
num_tests = 18500
plot_enrichment(enrichment, num_tests, plot_path, p_columns=["p_lof", "p_func"])
# remove the position column (which is only used to be able to locate the
# gene's position on a chromosome on a Manhattan plot).
del enrichment["start_pos"]
return enrichment
def test_get_de_novo_counts_synonymous(self):
""" check the de novo counts for synonymous consequences
"""
self.expected["synonymous_snv"] = 1
self.de_novos["consequence"] = 'synonymous_variant'
computed = get_de_novo_counts(self.de_novos)
self.compare_tables(computed, self.expected)
def test_get_de_novo_counts_synonymous(self):
""" check the de novo counts for synonymous consequences
"""
self.expected["synonymous_snv"] = 1
self.de_novos["consequence"] = 'synonymous_variant'
computed = get_de_novo_counts(self.de_novos)
self.compare_tables(computed, self.expected)
def test_get_de_novo_counts_type_included(self):
""" check the de novo counts when a type column is included
"""
# set the variant type before counting
self.de_novos["type"] = "snv"
self.expected["missense_snv"] = 1
computed = get_de_novo_counts(self.de_novos)
self.compare_tables(computed, self.expected)
def test_get_de_novo_counts_type_included(self):
""" check the de novo counts when a type column is included
"""
# set the variant type before counting
self.de_novos["type"] = "snv"
self.expected["missense_snv"] = 1
computed = get_de_novo_counts(self.de_novos)
self.compare_tables(computed, self.expected)
def test_get_de_novo_counts_nonfunctional(self):
""" check the de novo counts for nonfunctional consequence types
"""
snv = ["transcript_ablation", "transcript_amplification", \
"incomplete_terminal_codon_variant", "stop_retained_variant", \
"mature_miRNA_variant", \
"5_prime_UTR_variant", "3_prime_UTR_variant", \
"non_coding_transcript_exon_variant", "intron_variant", \
"NMD_transcript_variant", "non_coding_transcript_variant", \
"upstream_gene_variant", "downstream_gene_variant", \
"TFBS_ablation", "TFBS_amplification", "TF_binding_site_variant", \
"regulatory_region_ablation", "regulatory_region_amplification", \
"feature_elongation", "regulatory_region_variant", \
"feature_truncation", "intergenic_variant"]
if pandas.__version__ < '0.18.0':
# a dataframe with only nonfunctional consequences should raise an
# error, due to difficulties in counting rows
for cq in snv:
self.de_novos["consequence"] = cq
with self.assertRaises(KeyError):
computed = get_de_novo_counts(self.de_novos)
else:
expected = DataFrame(columns=[
'hgnc', 'chrom', 'start_pos', 'lof_indel', 'lof_snv',
'missense_indel', 'missense_snv', 'synonymous_snv'
])
for cq in snv:
self.de_novos["consequence"] = cq
computed = get_de_novo_counts(self.de_novos)
self.compare_tables(computed, expected)
# add a functional variant to the dataframe, so that the counting
# doesn't raise an error
new_row = self.de_novos.copy()
new_row["consequence"] = "missense_variant"
self.de_novos = self.de_novos.append(new_row, ignore_index=True)
self.expected["missense_snv"] = 1
for cq in snv:
self.de_novos.loc[0, "consequence"] = cq
computed = get_de_novo_counts(self.de_novos)
self.compare_tables(computed, self.expected)
def test_get_de_novo_counts_nonfunctional(self):
""" check the de novo counts for nonfunctional consequence types
"""
snv = ["transcript_ablation", "transcript_amplification", \
"incomplete_terminal_codon_variant", "stop_retained_variant", \
"mature_miRNA_variant", \
"5_prime_UTR_variant", "3_prime_UTR_variant", \
"non_coding_transcript_exon_variant", "intron_variant", \
"NMD_transcript_variant", "non_coding_transcript_variant", \
"upstream_gene_variant", "downstream_gene_variant", \
"TFBS_ablation", "TFBS_amplification", "TF_binding_site_variant", \
"regulatory_region_ablation", "regulatory_region_amplification", \
"feature_elongation", "regulatory_region_variant", \
"feature_truncation", "intergenic_variant"]
if pandas.__version__ < '0.18.0':
# a dataframe with only nonfunctional consequences should raise an
# error, due to difficulties in counting rows
for cq in snv:
self.de_novos["consequence"] = cq
with self.assertRaises(KeyError):
computed = get_de_novo_counts(self.de_novos)
else:
expected = DataFrame(columns=['hgnc', 'chrom', 'start_pos',
'lof_indel', 'lof_snv', 'missense_indel', 'missense_snv',
'synonymous_snv'])
for cq in snv:
self.de_novos["consequence"] = cq
computed = get_de_novo_counts(self.de_novos)
self.compare_tables(computed, expected)
# add a functional variant to the dataframe, so that the counting
# doesn't raise an error
new_row = self.de_novos.copy()
new_row["consequence"] = "missense_variant"
self.de_novos = self.de_novos.append(new_row, ignore_index=True)
self.expected["missense_snv"] = 1
for cq in snv:
self.de_novos.loc[0, "consequence"] = cq
computed = get_de_novo_counts(self.de_novos)
self.compare_tables(computed, self.expected)