catch rare product None cases #415

Author: oschwengers

bakta/features/cds.py

@@ -560,21 +560,22 @@ def revise_special_cases_annotated(data: dict, cdss: Sequence[dict]):
         for cds in seq_cdss:
             # look for supposedly truncated genes on rotated sequences
             if cds['start'] == 1  and  cds['strand'] == bc.STRAND_FORWARD  and  'truncated' in cds  and  cds['start_type'] == 'Edge'  and  cds['rbs_motif'] is None:
-                cds_gene_symbol = cds.get('gene', '-')
-                if RE_CHROM_ROTATION_GENE.fullmatch(cds_gene_symbol) or any(map(lambda x: x is not None, [RE_CHROM_ROTATION_GENE.fullmatch(term) for term in cds['product'].split()])):
+                cds_gene_symbol = cds.get('gene', '-')  or  '-'
+                cds_product = cds.get('product', '-')  or  '-'
+                if RE_CHROM_ROTATION_GENE.fullmatch(cds_gene_symbol) or any(map(lambda x: x is not None, [RE_CHROM_ROTATION_GENE.fullmatch(term) for term in cds_product.split()])):
                     # look for dnaA genes on rotated chromosome starts
                     cds.pop('truncated')
                     log.info(
                         'revise supposedly truncated dnaA gene on rotated chromosome start: seq=%s, start=%i, stop=%i, strand=%s, gene=%s, product=%s, nt=[%s..%s], aa=[%s..%s]',
-                        cds['sequence'], cds['start'], cds['stop'], cds['strand'], cds_gene_symbol, cds['product'], cds['nt'][:10], cds['nt'][-10:], cds['aa'][:10], cds['aa'][-10:]
+                        cds['sequence'], cds['start'], cds['stop'], cds['strand'], cds_gene_symbol, cds_product, cds['nt'][:10], cds['nt'][-10:], cds['aa'][:10], cds['aa'][-10:]
                     )
                     break
-                elif RE_PLASMID_ROTATION_GENE.fullmatch(cds_gene_symbol) or any(map(lambda x: x is not None, [RE_PLASMID_ROTATION_GENE.fullmatch(term) for term in cds['product'].split()])):
+                elif RE_PLASMID_ROTATION_GENE.fullmatch(cds_gene_symbol) or any(map(lambda x: x is not None, [RE_PLASMID_ROTATION_GENE.fullmatch(term) for term in cds_product.split()])):
                     # look for repABC|parAB genes on rotated plasmid starts
                     cds.pop('truncated')
                     log.info(
                         'revise supposedly truncated repABC/parAB gene on rotated plasmid start: seq=%s, start=%i, stop=%i, strand=%s, gene=%s, product=%s, nt=[%s..%s], aa=[%s..%s]',
-                        cds['sequence'], cds['start'], cds['stop'], cds['strand'], cds_gene_symbol, cds['product'], cds['nt'][:10], cds['nt'][-10:], cds['aa'][:10], cds['aa'][-10:]
+                        cds['sequence'], cds['start'], cds['stop'], cds['strand'], cds_gene_symbol, cds_product, cds['nt'][:10], cds['nt'][-10:], cds['aa'][:10], cds['aa'][-10:]
                     )
                     break