def get_cds_seqrecords(alchemist, values=[], nucleotide=False, verbose=False):
cds_list = parse_feature_data(alchemist, values=values)
seqrecords = []
genomes_dict = {}
for cds in cds_list:
if not cds.genome_id in genomes_dict.keys():
if verbose:
print(f"...Retrieving parent genome for {cds.id}...")
phage_id_obj = querying.get_column(alchemist.metadata,
"phage.PhageID")
phage_obj = phage_id_obj.table
parent_genome_query = querying.build_select(
alchemist.graph,
phage_obj,
where=\
phage_id_obj==cds.genome_id)
parent_genome_data = mysqldb_basic.first(alchemist.engine,
parent_genome_query)
parent_genome = mysqldb.parse_phage_table_data(parent_genome_data)
genomes_dict.update({cds.genome_id: parent_genome})
if verbose:
print(f"Converting {cds.id}...")
cds.genome_length = genomes_dict[cds.genome_id].length
cds.set_seqfeature()
record = cds_to_seqrecord(cds, genomes_dict[cds.genome_id])
seqrecords.append(record)
return seqrecords
def test_parse_phage_table_data_1(self):
"""Verify standard MySQL genome data is parsed correctly
from a data dictionary returned from a SQL query."""
data_dict = {
"PhageID": "L5",
"Accession": "ABC123",
"Name": "L5_Draft",
"HostGenus": "Mycobacterium",
"Sequence": "ATCG".encode("utf-8"),
"Length": 10,
"DateLastModified": constants.EMPTY_DATE,
"Notes": "abc".encode("utf-8"),
"GC": 12.12,
"Cluster": "A",
"Subcluster": "A2",
"Status": "final",
"RetrieveRecord": 1,
"AnnotationAuthor": 1
}
self.genome1 = \
mysqldb.parse_phage_table_data(data_dict, gnm_type="mysql")
with self.subTest():
self.assertEqual(self.genome1.id, "L5")
with self.subTest():
self.assertEqual(self.genome1.accession, "ABC123")
with self.subTest():
self.assertEqual(self.genome1.name, "L5_Draft")
with self.subTest():
self.assertEqual(self.genome1.host_genus, "Mycobacterium")
with self.subTest():
self.assertEqual(self.genome1.seq, "ATCG")
with self.subTest():
self.assertIsInstance(self.genome1.seq, Seq)
with self.subTest():
self.assertEqual(self.genome1.length, 10)
with self.subTest():
self.assertEqual(self.genome1.date, constants.EMPTY_DATE)
with self.subTest():
self.assertEqual(self.genome1.description, "abc")
with self.subTest():
self.assertEqual(self.genome1.gc, 12.12)
with self.subTest():
self.assertEqual(self.genome1.cluster, "A")
with self.subTest():
self.assertEqual(self.genome1.subcluster, "A2")
with self.subTest():
self.assertEqual(self.genome1.annotation_status, "final")
with self.subTest():
self.assertEqual(self.genome1.retrieve_record, 1)
with self.subTest():
self.assertEqual(self.genome1.annotation_author, 1)
with self.subTest():
self.assertEqual(self.genome1.translation_table, 11)
with self.subTest():
self.assertEqual(self.genome1.type, "mysql")
def test_parse_phage_table_data_2(self):
"""Verify truncated MySQL genome data is parsed correctly
from a data dictionary returned from a SQL query."""
data_dict = {"PhageID": "L5"}
self.genome1 = mysqldb.parse_phage_table_data(data_dict)
with self.subTest():
self.assertEqual(self.genome1.id, "L5")
with self.subTest():
self.assertEqual(self.genome1.name, "")
with self.subTest():
self.assertEqual(self.genome1.translation_table, 11)
with self.subTest():
self.assertEqual(self.genome1.type, "")