def check_prereqs() -> List[str]:
"Checks if all required files and applications are around"
_binary_extensions = ['.h3f', '.h3i', '.h3m', '.h3p']
failure_messages = []
for binary_name in ['hmmpfam2', 'hmmscan', 'hmmpress']:
if not path.locate_executable(binary_name):
failure_messages.append("Failed to locate file: %r" % binary_name)
# Get all HMM profile names from XML file
for profile in ["PKSI-KR.hmm2", "PKSI-KS_N.hmm2", "PKSI-KS_C.hmm2", "PKSI-AT.hmm2",
"PKSI-ACP.hmm2", "PKSI-DH.hmm2", "Thioesterase.hmm2", "PKSI-ER.hmm2",
"aa-activating.aroundLys.hmm2", "p450.hmm2"]:
full_hmm_path = path.get_full_path(__file__, "data", profile)
if path.locate_file(full_hmm_path) is None:
failure_messages.append("Failed to locate file: %s" % profile)
continue
if profile.endswith(".hmm2"):
continue
for ext in _binary_extensions:
binary = "{hmm}{ext}".format(hmm=full_hmm_path, ext=ext)
if not path.locate_file(binary):
result = subprocessing.run_hmmpress(full_hmm_path)
if not result.successful():
failure_messages.append("Failed to hmmpress {!r}: {!r}".format(profile, result.stderr))
# hmmpress generates _all_ binary files in one go, so stop the loop
break
binary_mtime = os.path.getmtime(binary)
hmm_mtime = os.path.getmtime(full_hmm_path)
if hmm_mtime < binary_mtime:
# generated file younger than hmm profile, do nothing
continue
try:
for filename in glob.glob("{}.h3?".format(full_hmm_path)):
logging.debug("removing outdated file %r", filename)
os.remove(filename)
except OSError as err:
failure_messages.append("Failed to remove outdated binary file for %s: %s" %
(profile, err))
break
result = subprocessing.run_hmmpress(full_hmm_path)
if not result.successful():
failure_messages.append("Failed to hmmpress %r: %r" % (profile, result.stderr))
failure_messages.append("HMM binary files outdated. %s (changed: %s) vs %s (changed: %s)" %
(profile, datetime.datetime.fromtimestamp(hmm_mtime),
binary, datetime.datetime.fromtimestamp(binary_mtime)))
# hmmpress generates _all_ binary files in one go, so stop the loop
break
return failure_messages
def check_prereqs() -> List[str]:
failure_messages = []
for binary_name, optional in [('hmmscan', False), ('hmmpress', False)]:
if path.locate_executable(binary_name) is None and not optional:
failure_messages.append("Failed to locate executable for %r" %
binary_name)
markov_models = [path.get_full_path(__file__, 'data', filename) for filename in [
'abmotifs.hmm', 'dockingdomains.hmm',
'ksdomains.hmm', 'nrpspksdomains.hmm']]
binary_extensions = ['.h3f', '.h3i', '.h3m', '.h3p']
for hmm in markov_models:
if path.locate_file(hmm) is None:
failure_messages.append("Failed to locate file %r" % hmm)
continue
for ext in binary_extensions:
binary = "{}{}".format(hmm, ext)
if path.locate_file(binary) is None:
result = subprocessing.run_hmmpress(hmm)
if not result.successful():
failure_messages.append('Failed to hmmpress {!r}: {}'.format(hmm, result.stderr))
break
return failure_messages
def ensure_database_pressed(filepath: str,
return_not_raise: bool = False) -> List[str]:
""" Ensures that the given HMMer database exists and that the hmmpress
generated files aren't out of date.
Arguments:
filepath: the path to the HMMer database
return_not_raise: whether to catch errors and return their messages as strings
Returns:
any encountered error messages, will never be populated without return_not_raise == True
"""
components = [
"{}{}".format(filepath, ext)
for ext in ['.h3f', '.h3i', '.h3m', '.h3p']
]
if path.is_outdated(components, filepath):
logging.info("%s components missing or obsolete, re-pressing database",
filepath)
result = subprocessing.run_hmmpress(filepath)
if not result.successful():
msg = "Failed to hmmpress {!r}: {}".format(filepath, result.stderr)
if not return_not_raise:
raise RuntimeError(msg)
return [msg]
return []
def check_prereqs() -> List[str]:
""" Check that prereqs are satisfied. hmmpress is only required if the
databases have not yet been generated.
"""
failure_messages = []
for binary_name, optional in [('hmmsearch', False), ('hmmpress', False)]:
if path.locate_executable(binary_name) is None and not optional:
failure_messages.append("Failed to locate executable for %r" %
binary_name)
profiles = None
# Check that hmmdetails.txt is readable and well-formatted
try:
profiles = get_signature_profiles()
except ValueError as err:
failure_messages.append(str(err))
# the path to the markov model
hmm = path.get_full_path(__file__, 'data', 'bgc_seeds.hmm')
hmm_files = [os.path.join("data", sig.hmm_file) for sig in profiles]
if path.locate_file(hmm) is None:
# try to generate file from all specified profiles in hmmdetails
try:
with open(hmm, 'w') as all_hmms_handle:
for hmm_file in hmm_files:
with open(path.get_full_path(__file__, hmm_file),
'r') as handle:
all_hmms_handle.write(handle.read())
except OSError:
failure_messages.append('Failed to generate file {!r}'.format(hmm))
# if previous steps have failed, the remainder will too, so don't try
if failure_messages:
return failure_messages
binary_extensions = ['.h3f', '.h3i', '.h3m', '.h3p']
for ext in binary_extensions:
binary = "{}{}".format(hmm, ext)
if path.locate_file(binary) is None:
result = run_hmmpress(hmm)
if not result.successful():
failure_messages.append('Failed to hmmpress {!r}: {}'.format(
hmm, result.stderr))
break
return failure_messages
def ensure_database_pressed(filepath: str,
return_not_raise: bool = False) -> List[str]:
""" Ensures that the given HMMer database exists and that the hmmpress
generated files aren't out of date.
Arguments:
filepath: the path to the HMMer database
return_not_raise: whether to catch errors and return their messages as strings
Returns:
any encountered error messages, will never be populated without return_not_raise == True
"""
try:
modified_time = os.path.getmtime(filepath)
except FileNotFoundError as err:
if not return_not_raise:
raise
return [str(err)]
components = [
"{}{}".format(filepath, ext)
for ext in ['.h3f', '.h3i', '.h3m', '.h3p']
]
outdated = False
for component in components:
if not path.locate_file(
component) or os.path.getmtime(component) < modified_time:
logging.info("%s does not exist or is out of date, hmmpressing %s",
component, filepath)
outdated = True
break
if outdated:
result = subprocessing.run_hmmpress(filepath)
if not result.successful():
msg = "Failed to hmmpress {!r}: {}".format(filepath, result.stderr)
if not return_not_raise:
raise RuntimeError(msg)
return [msg]
return []
def check_prereqs() -> List[str]:
"Check if all required applications are around"
failure_messages = []
for binary_name in ['muscle', 'hmmscan', 'hmmpress', 'fasttree', 'java']:
if path.locate_executable(binary_name) is None:
failure_messages.append("Failed to locate file: %r" % binary_name)
for hmm in ['smcogs.hmm']:
hmm = path.get_full_path(__file__, 'data', hmm)
if path.locate_file(hmm) is None:
failure_messages.append("Failed to locate file %r" % hmm)
continue
for ext in ['.h3f', '.h3i', '.h3m', '.h3p']:
binary = "%s%s" % (hmm, ext)
if path.locate_file(binary) is None:
# regenerate them
result = subprocessing.run_hmmpress(hmm)
if not result.successful():
failure_messages.append("Failed to hmmpress %s: %s" %
(hmm, result.stderr.rstrip()))
break
return failure_messages
def check_prereqs() -> List[str]:
""" Check the prerequisites.
hmmscan: domain detection
blastp: CLF and starter unit analysis
HMMs: t2pks.hmm
Returns:
a list of strings describing any errors, if they occurred
"""
failure_messages = []
for binary_name in ['hmmscan', 'blastp']:
if path.locate_executable(binary_name) is None:
failure_messages.append("Failed to locate file: %r" % binary_name)
for hmm in ['t2pks.hmm']:
hmm = path.get_full_path(__file__, 'data', hmm)
if path.locate_file(hmm) is None:
failure_messages.append("Failed to locate file %r" % hmm)
continue
for ext in ['.h3f', '.h3i', '.h3m', '.h3p']:
binary = "%s%s" % (hmm, ext)
if path.locate_file(binary) is None:
# regenerate them
result = subprocessing.run_hmmpress(hmm)
if not result.successful():
failure_messages.append("Failed to hmmpress %s: %s" %
(hmm, result.stderr.rstrip()))
break
for blastdb in ['KSIII', 'AT', 'LIG']:
for ext in ['.fasta', '.phr', '.pin', '.psq']:
dbfile = path.get_full_path(__file__, 'data', blastdb + ext)
if path.locate_file(dbfile) is None:
failure_messages.append("Failed to locate file %r" % dbfile)
continue
return failure_messages
