def ncbiDereplicatedGenomes(self, include_user_reps):
"""Get identifiers from the dereplicated set of NCBI genome.
Identifiers are return for NCBI representative genomes
and NCBI genomes without a representative.
Parameters
----------
include_user_reps : bool
Flag indicating if NCBI genomes assigned to a
User representative should be returned.
Returns
-------
list
List of database genome identifiers.
"""
try:
genome_mngr = GenomeManager(self.cur, self.currentUser)
ncbi_genomes_ids = genome_mngr.ncbiGenomeIds()
self.cur.execute(
"SELECT id " + "FROM metadata_taxonomy " +
"WHERE (gtdb_representative = 'TRUE' " +
"OR gtdb_genome_representative IS NULL) " + "AND id = ANY(%s)",
(ncbi_genomes_ids, ))
derep_genome_ids = [genome_id[0] for genome_id in self.cur]
if include_user_reps:
self.cur.execute(
"SELECT id " + "FROM metadata_taxonomy " +
"WHERE (gtdb_representative = 'FALSE' " +
"AND gtdb_genome_representative LIKE %s " +
"AND id = ANY(%s))", (
'U%',
ncbi_genomes_ids,
))
derep_genome_ids += [genome_id[0] for genome_id in self.cur]
except GenomeDatabaseError as e:
raise e
return derep_genome_ids
def ncbiDereplicatedGenomes(self, include_user_reps):
"""Get identifiers from the dereplicated set of NCBI genome.
Identifiers are return for NCBI representative genomes
and NCBI genomes without a representative.
Parameters
----------
include_user_reps : bool
Flag indicating if NCBI genomes assigned to a
User representative should be returned.
Returns
-------
list
List of database genome identifiers.
"""
try:
genome_mngr = GenomeManager(self.cur, self.currentUser)
ncbi_genomes_ids = genome_mngr.ncbiGenomeIds()
self.cur.execute("SELECT id " +
"FROM metadata_taxonomy " +
"WHERE (gtdb_representative = 'TRUE' " +
"OR gtdb_genome_representative IS NULL) " +
"AND id = ANY(%s)", (ncbi_genomes_ids,))
derep_genome_ids = [genome_id[0] for genome_id in self.cur]
if include_user_reps:
self.cur.execute("SELECT id " +
"FROM metadata_taxonomy " +
"WHERE (gtdb_representative = 'FALSE' " +
"AND gtdb_genome_representative LIKE %s " +
"AND id = ANY(%s))", ('U%', ncbi_genomes_ids,))
derep_genome_ids += [genome_id[0] for genome_id in self.cur]
except GenomeDatabaseError as e:
raise e
return derep_genome_ids
def ncbiRepresentativeGenomes(self):
"""Get genome identifiers for all NCBI representative genomes.
Returns
-------
list
List of database identifiers for NCBI representative genomes.
"""
try:
genome_mngr = GenomeManager(self.cur, self.currentUser)
ncbi_genomes_ids = genome_mngr.ncbiGenomeIds()
self.cur.execute("SELECT id " +
"FROM metadata_taxonomy " +
"WHERE gtdb_representative = 'TRUE' " +
"AND id = ANY(%s)", (ncbi_genomes_ids,))
rep_genome_ids = [genome_id[0] for genome_id in self.cur]
except GenomeDatabaseError as e:
raise e
return rep_genome_ids
def ncbiRepresentativeGenomes(self):
"""Get genome identifiers for all NCBI representative genomes.
Returns
-------
list
List of database identifiers for NCBI representative genomes.
"""
try:
genome_mngr = GenomeManager(self.cur, self.currentUser)
ncbi_genomes_ids = genome_mngr.ncbiGenomeIds()
self.cur.execute("SELECT id " +
"FROM metadata_taxonomy " +
"WHERE gtdb_representative = 'TRUE' " +
"AND id = ANY(%s)", (ncbi_genomes_ids,))
rep_genome_ids = [genome_id[0] for genome_id in self.cur]
except GenomeDatabaseError as e:
raise e
return rep_genome_ids
