def test_case_insensitive(self):
organism = '10090'
original_name = 'Pou4f1'
ncbi_id = 18996
gene_matcher = gene.GeneMatcher(organism)
gene_matcher.genes = [
'Pou4F1', 'pou4F1', 'POU4F1', 'pou4f1', original_name
]
gene_matcher.run_matcher()
self.assertEqual(
len([g.ncbi_id for g in gene_matcher.genes if g.ncbi_id]), 1)
gene_matcher = gene.GeneMatcher(organism, case_insensitive=True)
gene_matcher.genes = [
'Pou4F1', 'pou4F1', 'POU4F1', 'pou4f1', original_name
]
gene_matcher.run_matcher()
self.assertEqual(
len([g.ncbi_id for g in gene_matcher.genes if g.ncbi_id]), 5)
self.assertEqual(
set([g.ncbi_id for g in gene_matcher.genes if g.ncbi_id]).pop(),
ncbi_id)
def ncbi_info(taxid, genes, advance=None):
ensure_downloaded(gene.DOMAIN, gene.FILENAME, advance)
info = get_ncbi_info(taxid)
gene_matcher = gene.GeneMatcher(str(taxid))
gene_matcher.genes = genes
gene_matcher.run_matcher()
map_input_to_ensembl = OrderedDict(
{input_name: ''
for input_name in genes})
ret = []
for gene_obj in gene_matcher.get_known_genes():
if gene_obj.ncbi_id:
gi = info.get_gene_by_id(gene_obj.ncbi_id)
ensemble = ''
if 'Ensembl' in gi.db_refs:
ensemble = gi.db_refs['Ensembl']
map_input_to_ensembl[gene_obj.input_name] = ensemble
ret.append((gene_obj.input_name, [
str(gi.gene_id),
str(ensemble), gi.symbol + " (%s)" % gene_obj.input_name
if gene_obj.input_name != gi.symbol else gi.symbol,
gi.locus_tag or "", gi.chromosome or "", gi.description or "",
gi.synonyms, gi.symbol_from_nomenclature_authority or ""
]))
else:
ret.append(None)
return map_input_to_ensembl, ret
def _update_gene_matcher(self):
self._gene_names_from_table()
if not self.gene_matcher:
self.gene_matcher = gene.GeneMatcher(self._get_selected_organism())
self.gene_matcher.genes = self.input_genes
self.gene_matcher.organism = self._get_selected_organism()
def test_multiple_hits_scenario(self):
input_gene_name = 'HB1'
organism = '9606'
gene_matcher = gene.GeneMatcher(organism)
gene_matcher.genes = [input_gene_name]
gene_matcher.run_matcher()
result = gene_matcher.genes[0]
self.assertEqual(result.input_name, input_gene_name)
self.assertEqual(result.type_of_match, None)
self.assertEqual(result.ncbi_id, None)
self.assertGreater(len(result.possible_hits), 0)
def test_symbol_match_scenario(self):
input_gene_name = 'SCN5A'
ncbi_id = 6331
organism = '9606'
gene_matcher = gene.GeneMatcher(organism)
gene_matcher.genes = [input_gene_name]
gene_matcher.run_matcher()
result = gene_matcher.genes[0]
self.assertEqual(result.input_name, input_gene_name)
self.assertEqual(result.type_of_match, gene._symbol)
self.assertEqual(result.ncbi_id, ncbi_id)
result.load_ncbi_info()
for tag in gene.GENE_INFO_TAGS:
self.assertIsNotNone(getattr(result, tag))
def handle_input(self, data):
if data:
self.input_data = data
self.gene_matcher = gene.GeneMatcher(self._get_selected_organism())
self.gene_column_combobox.clear()
self.column_candidates = [
attr for attr in data.domain.variables + data.domain.metas
if isinstance(attr, (StringVariable, DiscreteVariable))
]
for var in self.column_candidates:
self.gene_column_combobox.addItem(*attributeItem(var))
self.tax_id = str(data_hints.get_hint(self.input_data, TAX_ID))
self.use_attr_names = data_hints.get_hint(
self.input_data, GENE_NAME, default=self.use_attr_names)
self.gene_col_index = min(self.gene_col_index,
len(self.column_candidates) - 1)
if self.tax_id in self.organisms:
self.selected_organism = self.organisms.index(self.tax_id)
self.on_input_option_change()
def test_types(self):
with self.assertRaises(TypeError):
gene.GeneMatcher(9606)