def is_database_exists(self):
"""Checks pfam database and interacdome table data exist. Overwrites Pfam.is_database_exists"""
try:
Pfam.is_database_exists(self)
except ConfigError:
raise ConfigError("It seems you do not have the associated Pfam data required to use "
"InteracDome, please run 'anvi-setup-interacdome' to download it. Then "
"run this command again.")
def __init__(self, args, run=terminal.Run(), progress=terminal.Progress()):
self.run = run
self.progress = progress
self.run.warning("Anvi'o will use 'InteracDome' by Kobren and Singh (DOI: 10.1093/nar/gky1224) to attribute binding frequencies. "
"If you publish your findings, please do not forget to properly credit their work.", lc='green', header="CITATION")
A = lambda x, t: t(args.__dict__[x]) if x in args.__dict__ else None
null = lambda x: x
self.interacdome_data_dir = A('interacdome_data_dir', null) or constants.default_interacdome_data_path
self.information_content_cutoff = A('information_content_cutoff', null) or 4
self.min_binding_frequency = A('min_binding_frequency', null) or 0
self.min_hit_fraction = A('min_hit_fraction', null) or 0.8
self.interacdome_dataset = A('interacdome_dataset', null) or 'representable'
self.output_prefix = A('output_file_prefix', null)
self.just_do_it = A('just_do_it', null)
self.run.warning("", header='INITIALIZATION', lc='green')
self.run.info("Interacdome dataset used", self.interacdome_dataset)
self.run.info("Minimum hit fraction", self.min_hit_fraction)
self.hmm_filepath = os.path.join(self.interacdome_data_dir, 'Pfam-A.hmm')
# Init the InteracDome table
self.interacdome_table = InteracDomeTableData(kind=self.interacdome_dataset, interacdome_data_dir=self.interacdome_data_dir)
self.interacdome_table.load()
# Init the Pfam baseclass
args.hmmer_program = 'hmmsearch' # Force use of hmmsearch
args.pfam_data_dir = self.interacdome_data_dir
Pfam.__init__(self, args, run=self.run, progress=self.progress)
# Init contigs database
args = argparse.Namespace(contigs_db=self.contigs_db_path)
self.contigs_db = dbops.ContigsSuperclass(args)
self.potentially_remove_previous_interacdome_data()
# Init the HMM profile
self.hmms = pfam.HMMProfile(self.hmm_filepath)
# This dictionary is populated and cast as a dataframe. It contains all of the per-residue
# binding frequency information for each hit
self.bind_freq = {}
# This dictionary (eventual dataframe) is just like self.bind_freq, except has averaged
# binding frequencies for residue-ligand combos that have multiple contributing hits. It
# also drops all contributing match state information
self.avg_bind_freq = {}
# This is a modified version of self.avg_bind_freq that is compatible with the
# amino_acid_additional_data table structure, i.e.
# tables.amino_acid_additional_data_table_structure
self.amino_acid_additional_data = {}