taxoniumtools 2.1.23__py3-none-any.whl → 2.1.24__py3-none-any.whl

taxoniumtools/_version.py

@@ -28,7 +28,7 @@ version_tuple: VERSION_TUPLE
 commit_id: COMMIT_ID
 __commit_id__: COMMIT_ID
 
-__version__ = version = '2.1.23'
-__version_tuple__ = version_tuple = (2, 1, 23)
+__version__ = version = '2.1.24'
+__version_tuple__ = version_tuple = (2, 1, 24)
 
-__commit_id__ = commit_id = 'g61a6dfcab'
+__commit_id__ = commit_id = 'g9d251e3c4'

taxoniumtools/ushertools.py

@@ -73,22 +73,29 @@ def get_codon_table():
 codon_table = get_codon_table()
 
 
-def get_gene_name(cds):
+def get_gene_name(cds, gene_records):
     """Returns gene if available, otherwise locus tag"""
     if "gene" in cds.qualifiers:
         return cds.qualifiers["gene"][0]
     elif "locus_tag" in cds.qualifiers:
         return cds.qualifiers["locus_tag"][0]
     else:
+        for gene in gene_records:
+            if gene.location == cds.location:
+                if "gene" in gene.qualifiers:
+                    return gene.qualifiers["gene"][0]
+        if "product" in cds.qualifiers:
+            return cds.qualifiers["product"][0].replace(" ", "_")
         raise ValueError(f"No gene name or locus tag for {cds}")
 
 
-def get_genes_dict(cdses):
+def get_genes_dict(cdses, gene_records):
     genes = {}
     for cds in cdses:
 
-        genes[get_gene_name(cds)] = Gene(get_gene_name(cds), cds.strand,
-                                         cds.location.start, cds.location.end)
+        gene_name = get_gene_name(cds, gene_records)
+        genes[gene_name] = Gene(gene_name, cds.strand, cds.location.start,
+                                cds.location.end)
     return genes
 
 
@@ -333,17 +340,18 @@ class UsherMutationAnnotatedTree:
     def load_genbank_file(self, genbank_file):
         self.genbank = SeqIO.read(genbank_file, "genbank")
         self.cdses = [x for x in self.genbank.features if x.type == "CDS"]
+        gene_records = [x for x in self.genbank.features if x.type == "gene"]
         # Assert that there are no compound locations and that all strands are positive,
         # and that all CDS features are a multiple of 3
 
-        self.genes = get_genes_dict(self.cdses)
+        self.genes = get_genes_dict(self.cdses, gene_records)
 
         by_everything = defaultdict(lambda: defaultdict(dict))
         total_lengths = {}
 
         for feature in self.cdses:
 
-            gene_name = get_gene_name(feature)
+            gene_name = get_gene_name(feature, gene_records)
 
             nucleotide_counter = 0
             for part in feature.location.parts:
@@ -370,9 +378,11 @@ class UsherMutationAnnotatedTree:
                 codon_obj = Codon(feat_name, codon_index, codon_dict,
                                   self.genes[feat_name].strand)
 
-                assert len(codon_dict) % 3 == 0
-                for k, v in codon_dict.items():
-                    nuc_to_codon[v].append(codon_obj)
+                if len(codon_dict) % 3 == 0:
+                    for k, v in codon_dict.items():
+                        nuc_to_codon[v].append(codon_obj)
+                else:
+                    print(f"Skipping partial codon for feature {feat_name}")
 
         self.nuc_to_codon = nuc_to_codon
 

{taxoniumtools-2.1.23.dist-info → taxoniumtools-2.1.24.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: taxoniumtools
-Version: 2.1.23
+Version: 2.1.24
 Summary: Generate files for taxonium
 Home-page: https://github.com/theosanderson/taxonium
 Author: Theo Sanderson

{taxoniumtools-2.1.23.dist-info → taxoniumtools-2.1.24.dist-info}/RECORD

@@ -1,14 +1,14 @@
 taxoniumtools/__init__.py,sha256=iizd2XLvtBHVDz6j82ZWJ2eojVvbTmtJCWQGk5MtqhE,51
 taxoniumtools/__main__.py,sha256=0-mrc1_4NaJur-6SQXV6vGrJFFDuRRFvqeQvnlOA5l4,25
-taxoniumtools/_version.py,sha256=40pybiGtRvrK6Js32ScNhZEg5nfbx1765bqxQ0IamCg,714
+taxoniumtools/_version.py,sha256=c9729_qM883xccr_V43bsBBjoeG7fZF3f-dZHO91oaw,714
 taxoniumtools/newick_to_taxonium.py,sha256=05JMQgmYLEo4nZVGPB0afAlF2s93WE7Ka_jwd0kCw0A,6207
 taxoniumtools/parsimony_pb2.py,sha256=hE6jWYlgTpUbTS1uTjRKwVgPNO232QIu3LqjpJUlthU,15575
 taxoniumtools/usher_to_taxonium.py,sha256=fir7_p18f8CaFGX7H1tRTm5YSryMhmcuA_PlL3sY2m4,12295
-taxoniumtools/ushertools.py,sha256=zF_V0ySNgI5Wx-cjXbwigFRH0hIz6_s2LVLBk2UT06o,16435
+taxoniumtools/ushertools.py,sha256=Af5KApxNSjTT6NHpb2m8bM8jCjz1_DSKgsnUbiBCndo,16985
 taxoniumtools/utils.py,sha256=sNbadwK7CNLK-sQFf31Au8nG1rUmcrbGMKDSV7UJMJo,10477
 taxoniumtools/view_taxonium.py,sha256=LYWBhaXYrmOOrIHI2bBYhqNUw-HZRtYzdvMQb6ZombI,4640
-taxoniumtools-2.1.23.dist-info/METADATA,sha256=b7j_RAWsWTurIIlBwlGpjmYmzkvMIguMft3HFub__7w,2168
-taxoniumtools-2.1.23.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-taxoniumtools-2.1.23.dist-info/entry_points.txt,sha256=oIcDnAw81KjfG32-ePuBFvqS_wbLXWB8NmWo61o1LW0,183
-taxoniumtools-2.1.23.dist-info/top_level.txt,sha256=EEWGvODb1nR_CiPHUDxcfU4UzxtPnVN4yTywlT3Z0nE,14
-taxoniumtools-2.1.23.dist-info/RECORD,,
+taxoniumtools-2.1.24.dist-info/METADATA,sha256=e8Kl3sDSY9R_QlBQsgbwhyxTbTY3CiQqXq6lZzraMAY,2168
+taxoniumtools-2.1.24.dist-info/WHEEL,sha256=wUyA8OaulRlbfwMtmQsvNngGrxQHAvkKcvRmdizlJi0,92
+taxoniumtools-2.1.24.dist-info/entry_points.txt,sha256=oIcDnAw81KjfG32-ePuBFvqS_wbLXWB8NmWo61o1LW0,183
+taxoniumtools-2.1.24.dist-info/top_level.txt,sha256=EEWGvODb1nR_CiPHUDxcfU4UzxtPnVN4yTywlT3Z0nE,14
+taxoniumtools-2.1.24.dist-info/RECORD,,

{taxoniumtools-2.1.23.dist-info → taxoniumtools-2.1.24.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (80.9.0)
+Generator: setuptools (80.10.2)
 Root-Is-Purelib: true
 Tag: py3-none-any