def load_collections(self):
''' Load the collections_txt file, run some sanity checks, and figure out params for anvi_import_collection'''
collections = u.get_TAB_delimited_file_as_dictionary(self.collections_txt)
bad_groups = [g for g in collections if g not in self.group_names]
if bad_groups:
raise ConfigError('Some of the names in your collection_txt \
file ("%s") don\'t match the names of the \
groups in your samples_txt/fasta_txt. \
Here are the names that don\'t match: %s. \
And here are the group names we expect to find: \
%s' % (self.collections_txt, ', '.join(bad_groups), ', '.join(self.group_names)))
for group in collections:
filesnpaths.is_file_exists(collections[group]['collection_file'])
if not collections[group]['collection_name']:
raise ConfigError('You must specify a name for each collection in your collections_txt')
u.check_collection_name(collections[group]['collection_name'])
if collections[group].get('bins_info'):
filesnpaths.is_file_exists(collections[group]['bins_info'])
collections[group]['bins_info'] = '--bins-info %s' % collections[group]['bins_info']
else:
collections[group]['bins_info'] = ''
if collections[group].get('contigs_mode'):
collections[group]['contigs_mode'] = '--contigs-mode'
else:
collections[group]['contigs_mode'] = ''
self.collections = collections
def load_collections(self):
''' Load the collections_txt file, run some sanity checks, and figure out params for anvi_import_collection'''
collections = u.get_TAB_delimited_file_as_dictionary(self.collections_txt)
bad_groups = [g for g in collections if g not in self.group_names]
if bad_groups:
raise ConfigError('Some of the names in your collection_txt '
'file ("%s") don\'t match the names of the '
'groups in your samples_txt/fasta_txt. '
'Here are the names that don\'t match: %s. '
'And here are the group names we expect to find: '
'%s' % (self.collections_txt, ', '.join(bad_groups), ', '.join(self.group_names)))
for group in collections:
default_collection = collections[group].get('default_collection')
if default_collection:
# User can specify either a default collection OR collection from file
not_allowed_params = {'collection_name', 'collection_file', 'bins_info', 'contigs_mode'}
if any([collections[group][key] for key in not_allowed_params if key in collections[group].keys()]):
raise ConfigError('We encountered the following problem with your '
'collections_txt file ("%s"): you can choose '
'either using a default collection OR importing '
'a collection from a file. Yet, for "%s", you specificy '
'a default collection AND also specify some of the following '
'parameters: %s.' % (self.collections_txt, group, ", ".join(not_allowed_params)))
collections[group]['collection_name'] = 'DEFAULT'
collections[group]['contigs_mode'] = ''
else:
if not filesnpaths.is_file_exists(collections[group]['collection_file'], dont_raise=True):
raise ConfigError('We encountered the following problem with your '
'collections_txt file ("%s"): you did not specify '
'a valid collection file for "%s".' % (self.collections_txt, group))
if not collections[group]['collection_name']:
raise ConfigError('You must specify a name for each collection in your collections_txt')
u.check_collection_name(collections[group]['collection_name'])
if collections[group].get('bins_info'):
filesnpaths.is_file_exists(collections[group]['bins_info'])
collections[group]['bins_info'] = '--bins-info %s' % collections[group]['bins_info']
else:
collections[group]['bins_info'] = ''
if collections[group].get('contigs_mode'):
collections[group]['contigs_mode'] = '--contigs-mode'
else:
collections[group]['contigs_mode'] = ''
self.collections = collections
def load_collections(self):
''' Load the collections_txt file, run some sanity checks, and figure out params for anvi_import_collection'''
collections = u.get_TAB_delimited_file_as_dictionary(self.collections_txt)
bad_groups = [g for g in collections if g not in self.group_names]
if bad_groups:
raise ConfigError('Some of the names in your collection_txt \
file ("%s") don\'t match the names of the \
groups in your samples_txt/fasta_txt. \
Here are the names that don\'t match: %s. \
And here are the group names we expect to find: \
%s' % (self.collections_txt, ', '.join(bad_groups), ', '.join(self.group_names)))
for group in collections:
default_collection = collections[group].get('default_collection')
if default_collection:
# User can specify either a default collection OR collection from file
not_allowed_params = {'collection_name', 'collection_file', 'bins_info', 'contigs_mode'}
if any([collections[group][key] for key in not_allowed_params if key in collections[group].keys()]):
raise ConfigError('We encountered the following problem with your \
collections_txt file ("%s"): you can choose \
either using a default collection OR importing \
a collection from a file. Yet, for "%s", you specificy \
a default collection AND also specify some of the following \
parameters: %s.' % (self.collections_txt, group, ", ".join(not_allowed_params)))
collections[group]['collection_name'] = 'DEFAULT'
collections[group]['contigs_mode'] = ''
else:
if not filesnpaths.is_file_exists(collections[group]['collection_file'], dont_raise=True):
raise ConfigError('We encountered the following problem with your \
collections_txt file ("%s"): you did not specify \
a valid collection file for "%s".' % (self.collections_txt, group))
if not collections[group]['collection_name']:
raise ConfigError('You must specify a name for each collection in your collections_txt')
u.check_collection_name(collections[group]['collection_name'])
if collections[group].get('bins_info'):
filesnpaths.is_file_exists(collections[group]['bins_info'])
collections[group]['bins_info'] = '--bins-info %s' % collections[group]['bins_info']
else:
collections[group]['bins_info'] = ''
if collections[group].get('contigs_mode'):
collections[group]['contigs_mode'] = '--contigs-mode'
else:
collections[group]['contigs_mode'] = ''
self.collections = collections
def get_target_files_for_anvi_cluster_contigs(self):
import anvio.workflows as w
w.D(self.get_param_value_from_config(['anvi_cluster_contigs', 'run']))
if self.get_param_value_from_config(['anvi_cluster_contigs', 'run']) is not True:
# the user doesn't want to run this
return
w.D('hi2')
requested_drivers = self.get_param_value_from_config(['anvi_cluster_contigs', '--driver'])
if not requested_drivers:
raise ConfigError('You must specify a driver for anvi_cluster_contigs. '
'You specified \'"run": true\' for anvi_cluster_contigs, \
but provided no driver.')
if type(requested_drivers) != list:
requested_drivers = [requested_drivers]
incompatible_drivers = [d for d in requested_drivers if d not in list(driver_modules['binning'].keys())]
if incompatible_drivers:
raise ConfigError('The following drivers were listed in the config file for rule anvi_cluster_contigs '
'but they are not familiar to anvi\'o: %s' % ', '.join(incompatible_drivers))
# TODO: we should make sure drivers are installed. Maybe Ozcan or Meren are willing to do this?
for d in driver_modules['binning'].keys():
# let's make sure no parameters were set for a driver that was not listed
additional_parameters = self.get_param_value_from_config(['anvi_cluster_contigs', self.get_param_name_for_binning_driver(d)])
if additional_parameters:
if d not in requested_drivers:
raise ConfigError('You set the following parameters: "%s" for %s, but you did not '
'specify it as one of the drivers that should be used by anvi_cluster_contigs. '
'In order to reduce room for mistakes, we don\'t allow this.' % (additional_parameters, d))
collection_name = self.get_param_value_from_config(['anvi_cluster_contigs', '--collection-name'])
if '{driver}' not in collection_name:
if len(requested_drivers) > 1:
raise ConfigError('When using multiple binning algorithms, the --collection-name '
'for rule anvi_cluster_contigs must contain '
'the key word "{driver}" (including those curly brackets). '
'It appears you changed the --collection-name, which by '
'default is simply "{driver}" to: "%s". That\'s fine, \
but only as long as you have "{driver}" appear somewhere in \
the name, because otherwise the collections made by different \
binning algorithms would try to override each other, since they \
would all end up having the same name' % collection_name)
# if the key word '{driver}' is not in the collection name then it is a static collection name
example_collection_name = collection_name
else:
# if the key word '{driver}' IS in the collection name, then let's take one
# driver as example when we check if the collection name is valid.
example_collection_name = collection_name.format(driver=requested_drivers[0])
try:
u.check_collection_name(example_collection_name)
except ConfigError as e:
raise ConfigError('%s is not an acceptable collection name for anvi_cluster_contigs. '
'We tried it for one of the drivers you requested and this is the '
'error we got: %s' % (collection_name, e))
groups_of_size_one = []
target_files = []
for d in requested_drivers:
for g in self.group_names:
if self.group_sizes[g] > 1:
# add flag files of binning to the target files
# only if the group is larger than 1 (because otherwise, there is no merged profile)
target_files.append(self.get_flag_file_for_binning_driver(g, d))
else:
groups_of_size_one.append(g)
if groups_of_size_one:
run.warning('You requested to run anvi_cluster_contigs, but it will not run for '
'%s, since there is only one sample in this group, and hence there '
'will be no merged profile, which is required for anvi-cluster-contigs.' % ', '.join(groups_of_size_one))
return target_files