Attempt to add support for CTD abbreviations

Refs pombase/canto#2807

ctd_support.py

@@ -9,6 +9,8 @@
 ctd_mutation_regex = f'{aa}\d+{aa}(?:{ctd_repetition_regex})?'
 ctd_deletion_regex = f'(?:delta|\u0394|∆)(?:{ctd_repetition_regex})?'
 
+ctd_abbreviations = f'(?:CTD_S2|CTD_T4|CTD_S5|CTD_S7)'
+
 # This dictionary has keys that are the systematic IDs of the genes that have CTDs,
 # the value is a dictionary with:
 # - positions: a dictionary with keys that are the residues of the canonical repeat
@@ -136,4 +138,4 @@ def ctd_format_for_transvar(capture_groups: list[str], gene: dict) -> list[str]:
 
 def ctd_apply_syntax(ctd_substring: str):
     ctd_bits = re.findall(f'({ctd_mutation_regex}|{ctd_deletion_regex})', ctd_substring)
-    return 'CTD-' + ','.join(ctd_bits)
+    return 'CTD-' + ','.join(ctd_bits)

grammar.py

@@ -1,6 +1,6 @@
 import copy
 from ctd_support import ctd_apply_syntax, ctd_further_check, ctd_format_for_transvar
-from ctd_support import ctd_mutation_regex, ctd_deletion_regex
+from ctd_support import ctd_mutation_regex, ctd_deletion_regex, ctd_abbreviations
 from grammar_funs_and_vars import aa, nt, num, multi_nt_regex, multi_nt_apply_syntax, multi_nt_check_sequence, \
     multi_aa_regex, multi_aa_apply_syntax, multi_aa_check_sequence, \
     multi_aa_format_for_transvar, insertion_aa_format_for_transvar, multi_nt_format_for_transvar, \
@@ -145,7 +145,7 @@
     {
         'type': 'amino_acid_deletion_and_mutation',
         'rule_name': 'CTD',
-        'regex': f'(CTD-(?:(?:{ctd_mutation_regex}|{ctd_deletion_regex}),?\s?)+)$',
+        'regex': f'(CTD-(?:(?:{ctd_mutation_regex}|{ctd_deletion_regex}),?\s?)+|{ctd_abbreviations})$',
         'apply_syntax': lambda g: ctd_apply_syntax(g[0]),
         'further_check': ctd_further_check,
         'format_for_transvar': ctd_format_for_transvar

test_syntax_rules.py

@@ -80,6 +80,10 @@ def test_class_methods_CTD(self):
         groups = syntax_rule.get_groups('CTD-delta(r1-r12-2),Y1F(r1-r12),Y2F', genome['SPBC28F2.12'])
         self.assertEqual(groups, ('CTD-delta(r1-r12-2),Y1F(r1-r12),Y2F', ))
 
+        syntax_rule = find_rule(syntax_rules, 'amino_acid_deletion_and_mutation', 'CTD')
+        groups = syntax_rule.get_groups('CTD_S2', genome['SPBC28F2.12'])
+        self.assertEqual(groups, ('CTD_S2', ))
+
     def test_class_methods_nucleotides(self):
 
         syntax_rules = [SyntaxRule.parse_obj(r) for r in nucleotide_grammar]