def process(self, aa_sequences_file_path=None):
if self.search_with not in self.available_db_search_program_targets:
raise ConfigError("Anvi'o understands that you want to use '%s' to search for COGs, however, there is no\
database formatted under the COGs data directory for that program :/ You may need to\
re-run the COGs setup, UNLESS, you set up your COG data directory somewhere else than what\
anvi'o attempts to use at the moment ('%s'). If that is the case, this may be the best\
time to point the right directory using the --cog-data-dir parameter." % \
(self.search_with, self.COG_data_dir))
if not aa_sequences_file_path and not self.contigs_db_path:
raise ConfigError(
"You either need to provide an anvi'o contigs database path, or a FASTA file for AA\
sequences")
if aa_sequences_file_path and self.contigs_db_path:
raise ConfigError(
"You can't provide both an AA sequences file and a contigs database. Choose one!"
)
if self.contigs_db_path:
dbops.is_contigs_db(self.contigs_db_path)
if not self.temp_dir_path:
self.temp_dir_path = filesnpaths.get_temp_directory_path()
self.remove_temp_dir_path = True
else:
filesnpaths.is_file_exists(self.temp_dir_path)
filesnpaths.is_output_dir_writable(self.temp_dir_path)
self.run.warning(
"Because you set the temporary directory path by hand, anvi'o will not remove its content\
when it is done. But she certainly hopes that you will clean those files later."
)
self.remove_temp_dir_path = False
self.run.info('COG data directory', self.COG_data_dir)
self.run.info('Directory to store temporary files', self.temp_dir_path)
self.run.info('Directory will be removed after the run',
self.remove_temp_dir_path)
if not aa_sequences_file_path:
aa_sequences_file_path = dbops.export_aa_sequences_from_contigs_db(
self.contigs_db_path, J(self.temp_dir_path, 'aa_sequences.fa'))
# do the search
search_results_tabular = self.search_factory[self.search_with](
aa_sequences_file_path)
# convert the output to a hits dict
self.hits = utils.get_BLAST_tabular_output_as_dict(
search_results_tabular,
target_id_parser_func=lambda x: x.split('|')[1])
# store hits into the contigs database
self.store_hits_into_contigs_db()
if self.remove_temp_dir_path:
shutil.rmtree(self.temp_dir_path)
def process(self, aa_sequences_file_path=None):
if self.search_with not in self.available_search_methods:
raise ConfigError("Let us start by making it clear that we probably like '%s' as much as you do, but it doesn't\
seem to be available on your system OR recognized by the COGs class since anvi'o couldn't\
find it among the available search methods. You probably need to try something else :/" \
% self.search_with)
if self.search_with not in self.available_db_search_program_targets:
raise ConfigError("Anvi'o understands that you want to use '%s' to search for COGs, however, there is no\
database formatted under the COGs data directory for that program :/ You may need to\
re-run the COGs setup, UNLESS, you set up your COG data directory somewhere else than what\
anvi'o attempts to use at the moment ('%s'). If that is the case, this may be the best\
time to point the right directory using the --cog-data-dir parameter, or the environmental\
variable 'ANVIO_COG_DATA_DIR'." % (self.search_with, self.COG_data_dir))
if not aa_sequences_file_path and not self.contigs_db_path:
raise ConfigError("You either need to provide an anvi'o contigs database path, or a FASTA file for AA\
sequences")
if aa_sequences_file_path and self.contigs_db_path:
raise ConfigError("You can't provide both an AA sequences file and a contigs database. Choose one!")
if self.contigs_db_path:
utils.is_contigs_db(self.contigs_db_path)
if not self.temp_dir_path:
self.temp_dir_path = filesnpaths.get_temp_directory_path()
self.remove_temp_dir_path = True
else:
filesnpaths.is_file_exists(self.temp_dir_path)
filesnpaths.is_output_dir_writable(self.temp_dir_path)
self.run.warning("Because you set the temporary directory path by hand, anvi'o will not remove its content\
when it is done. But she certainly hopes that you will clean those files later.")
self.remove_temp_dir_path = False
self.run.info('COG data directory', self.COG_data_dir)
self.run.info('Searching with', self.search_with)
self.run.info('Directory to store temporary files', self.temp_dir_path)
self.run.info('Directory will be removed after the run', self.remove_temp_dir_path)
if not aa_sequences_file_path:
aa_sequences_file_path = dbops.export_aa_sequences_from_contigs_db(self.contigs_db_path, J(self.temp_dir_path, 'aa_sequences.fa'))
# do the search
search_results_tabular = self.search_methods_factory[self.search_with](aa_sequences_file_path)
# convert the output to a hits dict
self.hits = utils.get_BLAST_tabular_output_as_dict(search_results_tabular, target_id_parser_func=lambda x: x.split('|')[1])
# store hits into the contigs database
self.store_hits_into_contigs_db()
if self.remove_temp_dir_path:
shutil.rmtree(self.temp_dir_path)
def process(self, aa_sequences_file_path=None):
if self.search_with not in self.available_search_methods:
raise ConfigError("Let us start by making it clear that we probably like '%s' as much as you do, but it doesn't\
seem to be available on your system OR recognized by the COGs class since anvi'o couldn't\
find it among the available search methods. You probably need to try something else :/" \
% self.search_with)
if self.search_with not in self.available_db_search_program_targets:
raise ConfigError(
"Anvi'o understands that you want to use '%s' to search for COGs, however, there is no\
database formatted under the COGs data directory for that program :/ You may need to\
re-run the COGs setup (anvi-setup-ncbi-cogs), UNLESS, you set up your COG data directory \
somewhere else than what anvi'o attempts to use at the moment ('%s'). If that is the case, \
this may be the best time to point the right directory using the --cog-data-dir parameter, \
or the environmental variable 'ANVIO_COG_DATA_DIR'."
% (self.search_with, self.COG_data_dir))
if not aa_sequences_file_path and not self.contigs_db_path:
raise ConfigError(
"You either need to provide an anvi'o contigs database path, or a FASTA file for AA\
sequences")
if aa_sequences_file_path and self.contigs_db_path:
raise ConfigError(
"You can't provide both an AA sequences file and a contigs database. Choose one!"
)
if self.contigs_db_path:
utils.is_contigs_db(self.contigs_db_path)
if not self.temp_dir_path:
self.temp_dir_path = filesnpaths.get_temp_directory_path()
self.remove_temp_dir_path = True
else:
filesnpaths.is_file_exists(self.temp_dir_path)
filesnpaths.is_output_dir_writable(self.temp_dir_path)
self.run.warning(
"Because you set the temporary directory path by hand, anvi'o will not remove its content\
when it is done. But she certainly hopes that you will clean those files later."
)
self.remove_temp_dir_path = False
self.run.info('COG data directory', self.COG_data_dir)
self.run.info('Searching with', self.search_with)
self.run.info('Directory to store temporary files', self.temp_dir_path)
self.run.info('Directory will be removed after the run',
self.remove_temp_dir_path)
if not aa_sequences_file_path:
aa_sequences_file_path = dbops.export_aa_sequences_from_contigs_db(
self.contigs_db_path, J(self.temp_dir_path, 'aa_sequences.fa'))
# do the search
search_results_tabular = self.search_methods_factory[self.search_with](
aa_sequences_file_path)
# convert the output to a hits dict
self.hits = utils.get_BLAST_tabular_output_as_dict(
search_results_tabular,
target_id_parser_func=lambda x: x.split('|')[1])
# store hits into the contigs database
self.store_hits_into_contigs_db()
if self.remove_temp_dir_path:
shutil.rmtree(self.temp_dir_path)