def test_find_latest(self):
with tempfile.TemporaryDirectory(prefix="aS.pfamdbtest") as temp_db_layout:
os.makedirs(os.path.join(temp_db_layout, "pfam", "30.7invalid"))
os.makedirs(os.path.join(temp_db_layout, "pfam", "invalid30.7"))
os.makedirs(os.path.join(temp_db_layout, "pfam", "irrelevant"))
with self.assertRaisesRegex(Exception, "No matching PFAM database in location " + temp_db_layout):
pfamdb.find_latest_database_version(temp_db_layout)
os.makedirs(os.path.join(temp_db_layout, "pfam", "31.0"))
assert pfamdb.find_latest_database_version(temp_db_layout) == "31.0"
os.makedirs(os.path.join(temp_db_layout, "pfam", "31.2"))
assert pfamdb.find_latest_database_version(temp_db_layout) == "31.2"
os.makedirs(os.path.join(temp_db_layout, "pfam", "30.7"))
assert pfamdb.find_latest_database_version(temp_db_layout) == "31.2"
def run_on_record(record: Record, results: Optional[hmmer.HmmerResults],
options: ConfigType) -> hmmer.HmmerResults:
""" Run hmmsearch against PFAM for all CDS features within the record """
if options.clusterhmmer_pfamdb_version == "latest":
database_version = pfamdb.find_latest_database_version(
options.database_dir)
else:
database_version = options.clusterhmmer_pfamdb_version
if results:
previous_db = pfamdb.get_db_version_from_path(results.database)
# same version requested, so reuse the results
if database_version == previous_db:
return results
logging.debug("Replacing clusterhmmer results from %s with %s",
previous_db, database_version)
logging.info('Running cluster PFAM search')
features = []
for region in record.get_regions():
features.extend(list(region.cds_children))
database = os.path.join(options.database_dir, 'pfam', database_version,
'Pfam-A.hmm')
return hmmer.run_hmmer(record, features, MAX_EVALUE, MIN_SCORE, database,
"clusterhmmer")
def run_on_record(record: Record, results: Optional[hmmer.HmmerResults],
options: ConfigType) -> hmmer.HmmerResults:
""" Run hmmsearch against PFAM for all CDS features within the record """
if options.fullhmmer_pfamdb_version == "latest":
database_version = pfamdb.find_latest_database_version(
options.database_dir)
else:
database_version = options.fullhmmer_pfamdb_version
if results:
previous_db = pfamdb.get_db_version_from_path(results.database)
# same version requested, so reuse the results
if database_version == previous_db:
return results
else:
logging.debug("Replacing fullhmmer results from %s with %s",
previous_db, database_version)
logging.info('Running whole-genome PFAM search')
database = os.path.join(options.database_dir, 'pfam', database_version,
'Pfam-A.hmm')
return hmmer.run_hmmer(record, record.get_cds_features(), MAX_EVALUE,
MIN_SCORE, database, "fullhmmer")
def setUp(self):
self._old_max_evalue = cluster_hmmer.MAX_EVALUE
self._old_min_score = cluster_hmmer.MIN_SCORE
cluster_hmmer.MAX_EVALUE = 0.02
cluster_hmmer.MIN_SCORE = 1.
self.config = build_config([],
isolated=True,
modules=antismash.get_all_modules())
self.latest_pfam = pfamdb.find_latest_database_version(
self.config.database_dir)
self.tracer = TraceTracker()
self.file_list = [
'Pfam-A.hmm', 'Pfam-A.hmm.h3f', 'Pfam-A.hmm.h3i', 'Pfam-A.hmm.h3m',
'Pfam-A.hmm.h3p'
]
mock('antismash.common.path.locate_file',
returns_iter=self.file_list,
tracker=self.tracer)
mock('antismash.common.subprocessing.run_hmmscan', returns=[])
self.expected_trace = """Called antismash.common.path.locate_file(
'{0}/pfam/{1}/Pfam-A.hmm')
Called antismash.common.path.locate_file(
'{0}/pfam/{1}/Pfam-A.hmm.h3f')
Called antismash.common.path.locate_file(
'{0}/pfam/{1}/Pfam-A.hmm.h3i')
Called antismash.common.path.locate_file(
'{0}/pfam/{1}/Pfam-A.hmm.h3m')
Called antismash.common.path.locate_file(
'{0}/pfam/{1}/Pfam-A.hmm.h3p')""".format(self.config.database_dir,
self.latest_pfam)
def check_options(options: ConfigType) -> List[str]:
""" Check the requested PFAM database exists """
database_version = options.fullhmmer_pfamdb_version
pfam_dir = os.path.join(options.database_dir, "pfam")
if database_version == "latest":
database_version = pfamdb.find_latest_database_version(options.database_dir)
return pfamdb.check_db(os.path.join(pfam_dir, database_version))
def setUp(self):
self._old_max_evalue = full_hmmer.MAX_EVALUE
self._old_min_score = full_hmmer.MIN_SCORE
full_hmmer.MAX_EVALUE = 0.02
full_hmmer.MIN_SCORE = 1.
self.config = build_config([],
isolated=True,
modules=antismash.get_all_modules())
self.latest_pfam = pfamdb.find_latest_database_version(
self.config.database_dir)
def run_on_record(record, results, options) -> hmmer.HmmerResults:
""" Run hmmsearch against PFAM for all CDS features within the record """
if results:
return results
logging.info('Running whole-genome PFAM search')
if options.fullhmmer_pfamdb_version == "latest":
database_version = pfamdb.find_latest_database_version(
options.database_dir)
else:
database_version = options.fullhmmer_pfamdb_version
database = os.path.join(options.database_dir, 'pfam', database_version,
'Pfam-A.hmm')
return hmmer.run_hmmer(record, record.get_cds_features(), MAX_EVALUE,
MIN_SCORE, database, "fullhmmer")
def check_prereqs() -> List[str]:
""" Ensure at least one database exists and is valid """
failure_messages = []
for binary_name in ['hmmscan']:
if not path.locate_executable(binary_name):
failure_messages.append("Failed to locate executable: %r" % binary_name)
data_dir = get_config().database_dir
try:
version = pfamdb.find_latest_database_version(data_dir)
except ValueError as err:
failure_messages.append(str(err))
return failure_messages
data_path = os.path.join(data_dir, "pfam", version)
failure_messages.extend(pfamdb.check_db(data_path))
return failure_messages
def run_on_record(record, results, options) -> hmmer.HmmerResults:
""" Run hmmsearch against PFAM for all CDS features within the record """
if results:
return results
logging.info('Running cluster PFAM search')
if options.clusterhmmer_pfamdb_version == "latest":
database_version = pfamdb.find_latest_database_version(
options.database_dir)
else:
database_version = options.clusterhmmer_pfamdb_version
database = os.path.join(options.database_dir, 'pfam', database_version,
'Pfam-A.hmm')
features = []
for cluster in record.get_clusters():
features.extend(list(cluster.cds_children))
return hmmer.run_hmmer(record, features, MAX_EVALUE, MIN_SCORE, database,
"clusterhmmer")
