def resolve_gtdb_family(self, gid_anis, ncbi_sp, type_gids, cur_genomes):
"""Resolve by identifying genomes with a conflicting GTDB family assignment."""
genus = 'g__' + generic_name(ncbi_sp)
gtdb_genus_rep = cur_genomes.gtdb_type_species_of_genus(genus)
if not gtdb_genus_rep:
return False, {}
expected_gtdb_family = cur_genomes[gtdb_genus_rep].gtdb_taxa.family
untrustworthy_gids = {}
matched_family = 0
for gid in type_gids:
if cur_genomes[gid].gtdb_taxa.family == expected_gtdb_family:
matched_family += 1
else:
# genome is classified to a different GTDB family than
# expected for this species
untrustworthy_gids[
gid] = f'Conflicting GTDB family assignment of {cur_genomes[gid].gtdb_taxa.family}, expected {expected_gtdb_family}'
all_similar = self.check_strain_ani(gid_anis, untrustworthy_gids)
# conflict is resolved if remaining genomes pass ANI similarity test,
if all_similar and len(untrustworthy_gids) > 0 and matched_family > 0:
return True, untrustworthy_gids
return False, {}
def resolve_gtdb_genus(self, gid_anis, ncbi_sp, type_gids, cur_genomes):
"""Resolve by identifying genomes with a conflicting GTDB genus assignments."""
ncbi_genus = 'g__' + generic_name(ncbi_sp)
untrustworthy_gids = {}
matched_genus = 0
for gid in type_gids:
canonical_gtdb_genus = canonical_taxon(cur_genomes[gid].gtdb_taxa.genus)
if ncbi_genus == canonical_gtdb_genus:
matched_genus += 1
else:
untrustworthy_gids[gid] = f'Conflicting GTDB genus assignment of {cur_genomes[gid].gtdb_taxa.genus}, expected {ncbi_genus}'
all_similar = self.check_strain_ani(gid_anis, untrustworthy_gids)
if all_similar and len(untrustworthy_gids) > 0 and matched_genus > 0:
return True, untrustworthy_gids
return False, {}
def infer_epithet_map(self, gids_of_interest, mc_species, cur_genomes,
cur_clusters):
"""Infer mapping of NCBI epithet to GTDB epithet which may be different due to gender of genus."""
# **************************************
# This should be updated so it only includes valid transfers, and not
# results due to misclassifications at NCBI. For example, right now this
# code reports Enterobacter cancerogenus being transferred to Pantoea, but
# really this is just a misclassified NCBI genome.
# get species in GTDB genus
generic_rids = defaultdict(list)
for rid in cur_clusters:
if rid not in gids_of_interest:
continue
gtdb_generic = cur_genomes[rid].gtdb_taxa.genus.replace('g__', '')
if rid in mc_species:
gtdb_generic = generic_name(mc_species[rid])
generic_rids[gtdb_generic].append(rid)
# establish epithets that are nearly identical
# except for small change to suffix which is
# assumed to be due to a gender change
for gtdb_generic, rids in generic_rids.items():
ncbi_sp_epithet_list = defaultdict(list)
for rid in rids:
ncbi_species = cur_genomes[rid].ncbi_taxa.species
if ncbi_species == 's__':
continue
ncbi_generic = generic_name(ncbi_species)
ncbi_specific = specific_epithet(ncbi_species)
if rid in mc_species:
gtdb_species = mc_species[rid]
else:
gtdb_species = cur_genomes[rid].gtdb_taxa.species
gtdb_specific = canonical_taxon(specific_epithet(gtdb_species))
self.gtdb_ncbi_generic_map[gtdb_generic][gtdb_specific].append(
ncbi_generic)
if test_same_epithet(ncbi_specific, gtdb_specific):
ncbi_sp_epithet_list[ncbi_specific].append(gtdb_specific)
for ncbi_specific, gtdb_specific_list in ncbi_sp_epithet_list.items(
):
gtdb_specific_counter = Counter(gtdb_specific_list)
top_gtdb_specific, count = gtdb_specific_counter.most_common(
1)[0]
map_perc = count * 100.0 / len(gtdb_specific_list)
if map_perc >= 50:
self.sp_epithet_map[gtdb_generic][
ncbi_specific] = top_gtdb_specific
if map_perc != 100:
self.logger.warning(
'Imperfect suffix mapping between from {} {} to {} at {:.1f}%.'
.format(gtdb_generic, top_gtdb_specific, ncbi_specific,
count * 100.0 / len(gtdb_specific_list)))
def run(self, manual_taxonomy, cur_gtdb_metadata_file, uba_genome_paths,
qc_passed_file, ncbi_genbank_assembly_file,
untrustworthy_type_file, synonym_file, gtdb_type_strains_ledger,
sp_priority_ledger, genus_priority_ledger, dsmz_bacnames_file):
"""Finalize species names based on results of manual curation."""
# initialize species priority manager
sp_priority_mngr = SpeciesPriorityManager(sp_priority_ledger,
genus_priority_ledger,
dsmz_bacnames_file)
# identify species and genus names updated during manual curation
self.logger.info('Parsing manually curated taxonomy.')
mc_taxonomy = Taxonomy().read(manual_taxonomy, use_canonical_gid=True)
self.logger.info(' - read taxonomy for {:,} genomes.'.format(
len(mc_taxonomy)))
# create current GTDB genome sets
self.logger.info('Creating current GTDB genome set.')
cur_genomes = Genomes()
cur_genomes.load_from_metadata_file(
cur_gtdb_metadata_file,
gtdb_type_strains_ledger=gtdb_type_strains_ledger,
create_sp_clusters=False,
uba_genome_file=uba_genome_paths,
qc_passed_file=qc_passed_file,
ncbi_genbank_assembly_file=ncbi_genbank_assembly_file,
untrustworthy_type_ledger=untrustworthy_type_file)
self.logger.info(
f' ... current genome set contains {len(cur_genomes):,} genomes.')
# get all GTDB species represented by a type strain:
gtdb_type_species = set()
for rid in mc_taxonomy:
if cur_genomes[rid].is_effective_type_strain():
gtdb_type_species.add(mc_taxonomy[rid][Taxonomy.SPECIES_INDEX])
# establish appropriate species names for GTDB clusters with new representatives
self.logger.info(
'Identifying type strain genomes with incongruent GTDB species assignments.'
)
fout = open(
os.path.join(self.output_dir, 'type_strains_incongruencies.tsv'),
'w')
fout.write(
'Genome ID\tGTDB species\tNCBI species\tGTDB type strain\tNCBI type strain\tNCBI RefSeq note\n'
)
num_incongruent = 0
for rid, taxa in mc_taxonomy.items():
if cur_genomes[rid].is_effective_type_strain():
gtdb_sp = taxa[Taxonomy.SPECIES_INDEX]
gtdb_generic = generic_name(gtdb_sp)
ncbi_sp = cur_genomes[rid].ncbi_taxa.species
ncbi_generic = generic_name(ncbi_sp)
if ncbi_sp == 's__':
# NCBI taxonomy is sometimes behind the genome annotation pages,
# and do not have a species assignment even for type strain genome
continue
# check if genome is a valid genus transfer into a genus
# that already contains a species with the specific
# name which results in a polyphyletic suffix being required
# e.g. G002240355 is Prauserella marina at NCBI and is
# transferred into Saccharomonospora under the GTDB. However,
# Saccharomonospora marina already exists so this genome
# needs to be S. marina_A.
if (is_placeholder_taxon(gtdb_sp)
and gtdb_generic != ncbi_generic
and canonical_species(gtdb_sp) in gtdb_type_species):
continue
if not test_same_epithet(specific_epithet(gtdb_sp),
specific_epithet(ncbi_sp)):
num_incongruent += 1
fout.write('{}\t{}\t{}\t{}\t{}\t{}\n'.format(
rid, gtdb_sp, ncbi_sp,
cur_genomes[rid].is_gtdb_type_strain(),
cur_genomes[rid].is_ncbi_type_strain(),
cur_genomes[rid].excluded_from_refseq_note))
self.logger.info(
' - identified {:,} genomes with incongruent species assignments.'.
format(num_incongruent))
fout.close()