def test_get_children_stats(f1_trio, person_set_id, male, female, unspecified,
count):
psc = f1_trio.get_person_set_collection("phenotype")
gh = GenotypeHelper(f1_trio, psc)
children_stats = gh.get_children_stats(person_set_id)
assert len(children_stats) == count
assert children_stats["M"] == male
assert children_stats["F"] == female
assert children_stats["U"] == unspecified
children_stats = gh.get_children_stats(person_set_id)
assert len(children_stats) == count
def __init__(
self, dataset, enrichment_tool, gene_syms):
self.dataset = dataset
self.gene_syms = gene_syms
self.tool = enrichment_tool
self.results = None
enrichment_config = self.tool.config
assert enrichment_config is not None
effect_types = expand_effect_types(enrichment_config.effect_types)
self.person_set_collection = self.dataset.get_person_set_collection(
enrichment_config.selected_person_set_collections[0]
)
self.gh = GenotypeHelper(
self.dataset, self.person_set_collection,
effect_types=effect_types)
def test_enrichment_tool(f1_trio, f1_trio_enrichment_config,
f1_trio_coding_len_background):
variants = list(
f1_trio.query_variants(inheritance=str(Inheritance.denovo.name)))
event_counter = EventsCounter()
enrichment_tool = EnrichmentTool(f1_trio_enrichment_config,
f1_trio_coding_len_background,
event_counter)
psc = f1_trio.get_person_set_collection("phenotype")
gh = GenotypeHelper(f1_trio, psc)
# children_stats = gh.get_children_stats()
children_by_sex = gh.children_by_sex("phenotype1")
enrichment_events = enrichment_tool.calc(
["missense", "synonymous"],
["SAMD11", "PLEKHN1", "POGZ"],
variants,
children_by_sex,
)
assert len(enrichment_events["all"].events) == 2
assert enrichment_events["all"].events == [["SAMD11"], ["SAMD11"]]
assert enrichment_events["all"].expected == 2.0
assert enrichment_events["all"].pvalue == 1.0
assert len(enrichment_events["rec"].events) == 1
assert enrichment_events["rec"].events == [["SAMD11"]]
assert enrichment_events["rec"].expected == 1.0
assert enrichment_events["rec"].pvalue == 1.0
assert len(enrichment_events["male"].events) == 1
assert enrichment_events["male"].events == [["SAMD11"]]
assert enrichment_events["male"].expected == 1.0
assert enrichment_events["male"].pvalue == 1.0
assert len(enrichment_events["female"].events) == 1
assert enrichment_events["female"].events == [["SAMD11"]]
assert enrichment_events["female"].expected == 1.0
assert enrichment_events["female"].pvalue == 1.0
assert len(enrichment_events["unspecified"].events) == 0
assert enrichment_events["unspecified"].events == []
assert enrichment_events["unspecified"].expected is None
assert enrichment_events["unspecified"].pvalue is None
def test_events_counter(f1_trio):
variants = list(
f1_trio.query_variants(inheritance=str(Inheritance.denovo.name))
)
psc = f1_trio.get_person_set_collection("phenotype")
gh = GenotypeHelper(f1_trio, psc)
# children_stats = gh.get_children_stats()
children_by_sex = gh.children_by_sex("phenotype1")
effect_types = set(["missense", "synonymous"])
event_counter = EventsCounter()
print(children_by_sex)
events = event_counter.events(variants, children_by_sex, effect_types)
print(events)
assert len(events["all"].events) == 2
assert events["all"].events == [["SAMD11"], ["SAMD11"]]
assert len(events["rec"].events) == 1
assert events["rec"].events == [["SAMD11"]]
assert len(events["male"].events) == 1
assert events["male"].events == [["SAMD11"]]
assert len(events["female"].events) == 1
assert events["female"].events == [["SAMD11"]]
assert len(events["unspecified"].events) == 0
def test_calc_stats(f1_trio, f1_trio_coding_len_background):
variants = list(
f1_trio.query_variants(inheritance=str(Inheritance.denovo.name)))
for fv in variants:
print(80 * "-")
print(fv, fv.effects)
for fa in fv.alleles:
print("\t", fa, fa.effects)
event_counter = EventsCounter()
psc = f1_trio.get_person_set_collection("phenotype")
gh = GenotypeHelper(f1_trio, psc)
children_stats = gh.get_children_stats("phenotype1")
children_by_sex = gh.children_by_sex("phenotype1")
enrichment_events = event_counter.events(variants, children_by_sex,
set(["missense", "synonymous"]))
assert len(enrichment_events["all"].events) == 2
assert enrichment_events["all"].events == [["SAMD11"], ["SAMD11"]]
assert enrichment_events["all"].expected is None
assert enrichment_events["all"].pvalue is None
assert len(enrichment_events["rec"].events) == 1
assert enrichment_events["rec"].events == [["SAMD11"]]
assert enrichment_events["rec"].expected is None
assert enrichment_events["rec"].pvalue is None
assert len(enrichment_events["male"].events) == 1
assert enrichment_events["male"].events == [["SAMD11"]]
assert enrichment_events["male"].expected is None
assert enrichment_events["male"].pvalue is None
assert len(enrichment_events["female"].events) == 1
assert enrichment_events["female"].events == [["SAMD11"]]
assert enrichment_events["female"].expected is None
assert enrichment_events["female"].pvalue is None
assert len(enrichment_events["unspecified"].events) == 0
assert enrichment_events["unspecified"].events == []
assert enrichment_events["unspecified"].expected is None
assert enrichment_events["unspecified"].pvalue is None
ee = f1_trio_coding_len_background.calc_stats(
["missense", "synonymous"],
enrichment_events,
["SAMD11", "PLEKHN1", "POGZ"],
children_stats,
)
assert ee == enrichment_events
assert len(ee["all"].events) == 2
assert ee["all"].events == [
["SAMD11"],
["SAMD11"],
]
assert ee["all"].expected == 2.0
assert ee["all"].pvalue == 1.0
assert len(ee["rec"].events) == 1
assert ee["rec"].events == [["SAMD11"]]
assert ee["rec"].expected == 1.0
assert ee["rec"].pvalue == 1.0
assert len(ee["male"].events) == 1
assert ee["male"].events == [["SAMD11"]]
assert ee["male"].expected == 1.0
assert ee["male"].pvalue == 1.0
assert len(ee["female"].events) == 1
assert ee["female"].events == [["SAMD11"]]
assert ee["female"].expected == 1.0
assert ee["female"].pvalue == 1.0
assert len(ee["unspecified"].events) == 0
assert ee["unspecified"].events == []
assert ee["unspecified"].expected is None
assert ee["unspecified"].pvalue is None
def test_calc_stats(f1_trio, f1_trio_samocha_background):
variants = list(
f1_trio.query_variants(inheritance=str(Inheritance.denovo.name)))
event_counter = EventsCounter()
psc = f1_trio.get_person_set_collection("phenotype")
gh = GenotypeHelper(f1_trio, psc)
# children_stats = gh.get_children_stats()
children_by_sex = gh.children_by_sex("phenotype1")
enrichment_events = event_counter.events(variants, children_by_sex,
set(["missense", "synonymous"]))
assert len(enrichment_events["all"].events) == 2
assert enrichment_events["all"].events == [["SAMD11"], ["SAMD11"]]
assert enrichment_events["all"].expected is None
assert enrichment_events["all"].pvalue is None
assert len(enrichment_events["rec"].events) == 1
assert enrichment_events["rec"].events == [["SAMD11"]]
assert enrichment_events["rec"].expected is None
assert enrichment_events["rec"].pvalue is None
assert len(enrichment_events["male"].events) == 1
assert enrichment_events["male"].events == [["SAMD11"]]
assert enrichment_events["male"].expected is None
assert enrichment_events["male"].pvalue is None
assert len(enrichment_events["female"].events) == 1
assert enrichment_events["female"].events == [["SAMD11"]]
assert enrichment_events["female"].expected is None
assert enrichment_events["female"].pvalue is None
assert len(enrichment_events["unspecified"].events) == 0
assert enrichment_events["unspecified"].events == []
assert enrichment_events["unspecified"].expected is None
assert enrichment_events["unspecified"].pvalue is None
ee = f1_trio_samocha_background.calc_stats(
"missense",
enrichment_events,
["SAMD11", "PLEKHN1", "POGZ"],
children_by_sex,
)
assert ee == enrichment_events
assert len(ee["all"].events) == 2
assert ee["all"].events == [["SAMD11"], ["SAMD11"]]
assert ee["all"].expected == 30.0
assert ee["all"].pvalue == pytest.approx(9.002e-11)
assert len(ee["rec"].events) == 1
assert ee["rec"].events == [["SAMD11"]]
assert ee["rec"].expected == 15.0
assert ee["rec"].pvalue == pytest.approx(9.788e-6, rel=1e-3)
assert len(ee["male"].events) == 1
assert ee["male"].events == [["SAMD11"]]
assert ee["male"].expected == 15.0
assert ee["male"].pvalue == pytest.approx(9.788e-06, rel=1e-3)
assert len(ee["female"].events) == 1
assert ee["female"].events == [
["SAMD11"],
]
assert ee["female"].expected == 15.0
assert ee["female"].pvalue == pytest.approx(9.788e-06, rel=1e-3)
assert len(ee["unspecified"].events) == 0
assert ee["unspecified"].events == []
assert ee["unspecified"].expected is None
assert ee["unspecified"].pvalue is None
class EnrichmentBuilder(BaseEnrichmentBuilder):
def __init__(
self, dataset, enrichment_tool, gene_syms):
self.dataset = dataset
self.gene_syms = gene_syms
self.tool = enrichment_tool
self.results = None
enrichment_config = self.tool.config
assert enrichment_config is not None
effect_types = expand_effect_types(enrichment_config.effect_types)
self.person_set_collection = self.dataset.get_person_set_collection(
enrichment_config.selected_person_set_collections[0]
)
self.gh = GenotypeHelper(
self.dataset, self.person_set_collection,
effect_types=effect_types)
def build_people_group_selector(self, effect_types, person_set):
children_stats = self.gh.get_children_stats(person_set.id)
children_count = (
children_stats["M"] + children_stats["F"] + children_stats["U"]
)
if children_count <= 0:
return None
results = {}
for effect_type in effect_types:
enrichment_results = self.tool.calc(
effect_type,
self.gene_syms,
self.gh.get_denovo_variants(),
self.gh.children_by_sex(person_set.id),
)
results[effect_type] = enrichment_results
results["childrenStats"] = self.gh.get_children_stats(person_set.id)
results["selector"] = person_set.name
results["geneSymbols"] = list(self.gene_syms)
results["peopleGroupId"] = self.person_set_collection.id
results["peopleGroupValue"] = person_set.id
results["datasetId"] = self.dataset.study_id
return results
def _build_results(self):
results = []
enrichment_config = self.tool.config
assert enrichment_config is not None
effect_types = enrichment_config.effect_types
for person_set in self.person_set_collection.person_sets.values():
res = self.build_people_group_selector(effect_types, person_set)
if res:
results.append(res)
return results
def build(self):
results = self._build_results()
serializer = EnrichmentSerializer(self.tool.config, results)
results = serializer.serialize()
self.results = results
return self.results