def test_genome_to_seqrecord_2(self):
"""Verify that genome_to_seqrecord can correctly
populate seqrecord annotations"""
self.gnm.seq = Seq("ATA")
self.gnm.date = "2019"
self.gnm.accession = "gnm12345"
self.gnm.host_genus = "Mycobacterium"
self.gnm.id = "Trixie"
self.gnm.cluster = "A"
self.gnm.subcluster = "A2"
self.gnm.annotation_status = "1"
self.gnm.annotation_author = 1
self.gnm.retrieve_record = 1
record = flat_files.genome_to_seqrecord(self.gnm)
record_comments = record.annotations["comment"]
exp_source = "Mycobacterium phage Trixie"
exp_comment0 = "Cluster: A; Subcluster: A2"
exp_comment2 = "Annotation Status: 1; Annotation Author: 1"
exp_comment3 = "RetrieveRecord: 1"
with self.subTest():
self.assertEqual(record.annotations["date"], "2019")
with self.subTest():
self.assertEqual(record.annotations["source"], exp_source)
with self.subTest():
self.assertEqual(record_comments[0], exp_comment0)
with self.subTest():
self.assertEqual(record_comments[2], exp_comment2)
with self.subTest():
self.assertEqual(record_comments[3], exp_comment3)
def get_genome_seqrecords(alchemist, values=[], verbose=False):
genomes = mysqldb.parse_genome_data(alchemist.engine,
phage_id_list=values,
phage_query=PHAGE_QUERY,
gene_query=GENE_QUERY)
seqrecords = []
for gnm in genomes:
process_cds_features(gnm)
if verbose:
print(f"Converting {gnm.name}...")
seqrecords.append(flat_files.genome_to_seqrecord(gnm))
return seqrecords
def get_genome_seqrecords(alchemist, values, data_cache=None, verbose=False):
if data_cache is None:
data_cache = {}
seqrecords = []
for genome_id in values:
genome = data_cache.get(genome_id)
if genome is None:
genome = get_single_genome(alchemist, genome_id, get_features=True,
data_cache=data_cache)
seqrecord = flat_files.genome_to_seqrecord(genome)
flat_files.sort_seqrecord_features(seqrecord)
seqrecords.append(seqrecord)
return seqrecords
def build_id_record_map(alchemist, phageids):
id_record_map = {}
if not phageids:
return id_record_map
genomes = mysqldb.parse_genome_data(alchemist.engine,
phage_id_list=phageids,
phage_query=PHAGE_QUERY,
gene_query=GENE_QUERY,
trna_query=TRNA_QUERY,
tmrna_query=TMRNA_QUERY)
for genome in genomes:
record = flat_files.genome_to_seqrecord(genome)
id_record_map[record.id] = record
return id_record_map
def test_genome_to_seqrecord_1(self):
"""Verify that genome_to_seqrecord can initialize
a SeqRecord given the basic conditions"""
self.gnm.seq = Seq("ATA")
self.gnm.name = "Trixie"
self.gnm.id = "Trixie"
self.gnm.cds_features = []
self.gnm.host_genus = "Mycobacterium"
record = flat_files.genome_to_seqrecord(self.gnm)
exp_description = "Mycobacterium phage Trixie, complete genome"
with self.subTest():
self.assertEqual(record.name, "Trixie")
with self.subTest():
self.assertEqual(record.features, [])
with self.subTest():
self.assertEqual(record.description, exp_description)
with self.subTest():
self.assertEqual(record.seq, Seq("ATA"))
def execute_ffx_export(alchemist,
output_path,
file_format,
db_version,
table="phage",
values=[],
verbose=False):
if verbose:
print(f"Retrieving {data_name} data from {sql_handle.database}...")
if table == "phage":
genomes = mysqldb.parse_genome_data(alchemist.engine,
phage_id_list=values,
phage_query="SELECT * FROM phage",
gene_query="SELECT * FROM gene")
else:
raise ValueError
if verbose:
print(f"Converting {data_name} data to SeqRecord format...")
seqrecords = []
if table == "phage":
for gnm in genomes:
set_cds_seqfeatures(gnm)
if verbose:
print(f"Converting {gnm.name}...")
seqrecords.append(flat_files.genome_to_seqrecord(gnm))
if verbose:
print("Appending database version...")
for record in seqrecords:
append_database_version(record, db_version)
else:
raise ValueError
write_seqrecord(seqrecords, file_format, output_path, verbose=verbose)
def test_genome_to_seqrecord_4(self):
gnm = ""
with self.assertRaises(AttributeError):
record = flat_files.genome_to_seqrecord(gnm)
def test_genome_to_seqrecord_3(self):
gnm = None
with self.assertRaises(AssertionError):
record = flat_files.genome_to_seqrecord(gnm)