2424# bacteria. for example 'ar53' and 'bac120' mean that the GTDB phylogeny
2525# is based on 53 and 120 concatenated proteins (cp), respectively.
2626# If this changes we can set up a conditional statemnt below.
27- VERSION_MAP_DICT = {'226.0' : {'Archaea' : 'ar53' , 'Bacteria' : 'bac120' },
27+ VERSION_MAP_DICT = {'232.0' : {'Archaea' : 'ar53' , 'Bacteria' : 'bac120' },
28+ '226.0' : {'Archaea' : 'ar53' , 'Bacteria' : 'bac120' },
2829 '220.0' : {'Archaea' : 'ar53' , 'Bacteria' : 'bac120' },
2930 '214.1' : {'Archaea' : 'ar53' , 'Bacteria' : 'bac120' },
3031 '214.0' : {'Archaea' : 'ar53' , 'Bacteria' : 'bac120' },
3334
3435
3536def get_gtdb_data (
36- version : str = '226 .0' ,
37+ version : str = '232 .0' ,
3738 domain : str = 'Both' ,
3839 db_type : str = 'SpeciesReps' ,
3940 url_type : str = 'Primary'
@@ -49,7 +50,7 @@ def get_gtdb_data(
4950 return tax_q , seqs_q
5051
5152
52- def _assemble_queries (version = '226 .0' ,
53+ def _assemble_queries (version = '232 .0' ,
5354 db_type = 'SpeciesReps' ,
5455 domain = 'Both' ,
5556 url_type = 'Primary' ):
@@ -66,7 +67,7 @@ def _assemble_queries(version='226.0',
6667 # file names...
6768 # GTDB v220 started storing the ssu_reps FASTA files
6869 # as 'fna.gz' instead of their usual 'tar.gz'.
69- if version in ['220.0' , '226.0' ]:
70+ if version in ['220.0' , '226.0' , '232.0' ]:
7071 stype = 'fna'
7172 else :
7273 stype = 'tar'
0 commit comments