def parse_feature_data(alchemist, values=[], limit=8000):
"""Returns Cds objects containing data parsed from a MySQL database.
:param alchemist: A connected and fully built AlchemyHandler object.
:type alchemist: AlchemyHandler
:param values: List of GeneIDs upon which the query can be conditioned.
:type values: list[str]
"""
gene_table = querying.get_table(alchemist.metadata, "gene")
primary_key = list(gene_table.primary_key.columns)[0]
cds_data_columns = list(gene_table.c)
cds_data_query = querying.build_select(alchemist.graph, cds_data_columns)
cds_data = querying.execute(alchemist.engine,
cds_data_query,
in_column=primary_key,
values=values,
limit=limit)
cds_list = []
for data_dict in cds_data:
cds_ftr = mysqldb.parse_gene_table_data(data_dict)
cds_list.append(cds_ftr)
return cds_list
def test_parse_gene_table_data_1(self):
"""Verify standard MySQL CDS data is parsed correctly
from a data dictionary returned from a SQL query."""
data_dict = {
"GeneID": "L5_001",
"PhageID": "L5",
"Start": 10,
"Stop": 100,
"Parts": 1,
"Length": 1000,
"Name": "1",
"Translation": "AGGPT",
"Orientation": "F",
"Notes": "description".encode("utf-8"),
"DomainStatus": 1,
"PhamID": 1,
"LocusTag": "SEA_L5_001"
}
cds1 = mysqldb.parse_gene_table_data(data_dict)
with self.subTest():
self.assertEqual(cds1.id, "L5_001")
with self.subTest():
self.assertEqual(cds1.genome_id, "L5")
with self.subTest():
self.assertEqual(cds1.start, 10)
with self.subTest():
self.assertEqual(cds1.stop, 100)
with self.subTest():
self.assertEqual(cds1.parts, 1)
with self.subTest():
self.assertEqual(cds1.length, 1000)
with self.subTest():
self.assertEqual(cds1.name, "1")
with self.subTest():
self.assertEqual(cds1.type, "CDS")
with self.subTest():
self.assertEqual(cds1.translation, "AGGPT")
with self.subTest():
self.assertEqual(cds1.orientation, "F")
with self.subTest():
self.assertEqual(cds1.description, "description")
with self.subTest():
self.assertEqual(cds1.locus_tag, "SEA_L5_001")
with self.subTest():
self.assertEqual(cds1.translation_table, 11)
with self.subTest():
self.assertEqual(cds1.coordinate_format, "0_half_open")
with self.subTest():
self.assertEqual(cds1.domain_status, 1)
with self.subTest():
self.assertEqual(cds1.pham_id, 1)
def parse_cds_data_from_geneid(sql_handle, geneid_list):
if not geneid_list:
return []
query = (f"SELECT * FROM gene WHERE GeneID IN ('" + \
"','".join(geneid_list) + "')")
result_list = sql_handle.execute_query(query)
cds_list = []
for data_dict in result_list:
cds_list.append(mysqldb.parse_gene_table_data(data_dict))
return cds_list
args = sys.argv
args.insert(0, "blank_argument")
main(args)
def test_parse_gene_table_data_2(self):
"""Verify truncated MySQL CDS data is parsed correctly
from a data dictionary returned from a SQL query."""
data_dict = {"GeneID":"L5_001",
"PhageID":"L5",
"Start":10,
"LocusTag":None}
cds1 = mysqldb.parse_gene_table_data(data_dict)
with self.subTest():
self.assertEqual(cds1.id, "L5_001")
with self.subTest():
self.assertEqual(cds1.genome_id, "L5")
with self.subTest():
self.assertEqual(cds1.start, 10)
with self.subTest():
self.assertEqual(cds1.locus_tag, "")
with self.subTest():
self.assertEqual(cds1.translation_table, 11)