get-gtdb-data: 214 release is new default (#157)

Thank you doing this @nbokulich! I was unaware of the new GTDB release. I ran all the code manually as well as the test code. It all works beautifully. :-)
bokulich-lab · Apr 28, 2023 · ed0344d · ed0344d
1 parent 93f9ab0
commit ed0344d
Show file tree

Hide file tree

Showing 3 changed files with 16 additions and 5 deletions.
diff --git a/rescript/get_gtdb.py b/rescript/get_gtdb.py
@@ -21,11 +21,12 @@
 # bacteria. for example 'ar53' and 'bac120' mean that the GTDB phylogeny
 # is based on 53 and 120 concatenated proteins (cp), respectively.
 # If this changes we can set up a conditional statemnt below.
-VERSION_MAP_DICT = {'207': {'Archaea': 'ar53', 'Bacteria': 'bac120'},
+VERSION_MAP_DICT = {'214': {'Archaea': 'ar53', 'Bacteria': 'bac120'},
+                    '207': {'Archaea': 'ar53', 'Bacteria': 'bac120'},
                     '202': {'Archaea': 'ar122', 'Bacteria': 'bac120'}}
 
 
-def get_gtdb_data(ctx, version='207', domain='Both'):
+def get_gtdb_data(ctx, version='214', domain='Both'):
 
     ver_dom_dict = defaultdict(lambda: defaultdict(dict))
 

diff --git a/rescript/plugin_setup.py b/rescript/plugin_setup.py
@@ -904,7 +904,7 @@
     function=get_gtdb_data,
     inputs={},
     parameters={
-        'version': Str % Choices(['202', '207']),
+        'version': Str % Choices(['202', '207', '214']),
         'domain': Str % Choices(['Both', 'Bacteria', 'Archaea']),
         },
     outputs=[('gtdb_taxonomy', FeatureData[Taxonomy]),

diff --git a/rescript/tests/test_get_gtdb.py b/rescript/tests/test_get_gtdb.py
@@ -51,6 +51,10 @@ def test_assemble_queries(self):
         obs_seq_urls = [q_info[1] for q_info in queries['Sequence']]
 
         exp_tax_urls = [('https://data.gtdb.ecogenomic.org/releases/'
+                         'release214/214.0/ar53_taxonomy_r214.tsv.gz'),
+                        ('https://data.gtdb.ecogenomic.org/releases/'
+                         'release214/214.0/bac120_taxonomy_r214.tsv.gz'),
+                        ('https://data.gtdb.ecogenomic.org/releases/'
                          'release207/207.0/ar53_taxonomy_r207.tsv.gz'),
                         ('https://data.gtdb.ecogenomic.org/releases/'
                          'release207/207.0/bac120_taxonomy_r207.tsv.gz'),
@@ -59,6 +63,12 @@ def test_assemble_queries(self):
                         ('https://data.gtdb.ecogenomic.org/releases'
                          '/release202/202.0/bac120_taxonomy_r202.tsv.gz')]
         exp_seq_urls = [('https://data.gtdb.ecogenomic.org/releases'
+                         '/release214/214.0/genomic_files_reps/'
+                         'ar53_ssu_reps_r214.tar.gz'),
+                        ('https://data.gtdb.ecogenomic.org/releases/'
+                         'release214/214.0/genomic_files_reps/'
+                         'bac120_ssu_reps_r214.tar.gz'),
+                        ('https://data.gtdb.ecogenomic.org/releases'
                          '/release207/207.0/genomic_files_reps/'
                          'ar53_ssu_reps_r207.tar.gz'),
                         ('https://data.gtdb.ecogenomic.org/releases/'
@@ -91,11 +101,11 @@ def _makey_fakey_arch(faking_ignore_this):
         def _makey_fakey_bact(faking_ignore_this):
             return [self.bact_tax], [self.bact_seqs]
 
-        # default (both domains, version 207)
+        # default (both domains, version 214)
         with patch('rescript.get_gtdb._retrieve_data_from_gtdb',
                    new=_makey_fakey_both):
             res = rescript.actions.get_gtdb_data(
-                version='207', domain='Both')
+                version='214', domain='Both')
             self.assertEqual(len(res), 2)
             self.assertEqual(str(res[0].type), 'FeatureData[Taxonomy]')
             self.assertEqual(str(res[1].type), 'FeatureData[Sequence]')