Made constellation regex validation more flexible

flu-crew · Nov 6, 2023 · 0baf6c5 · 0baf6c5
1 parent 7936ce6
commit 0baf6c5
Show file tree

Hide file tree

Showing 2 changed files with 4 additions and 1 deletion.
diff --git a/octofludb/domain_flu.py b/octofludb/domain_flu.py
@@ -16,7 +16,7 @@
 p_internal_gene = p.regex("PB2|PB1|PA|NP") ^ p_ns ^ p_m
 
 p_segment = p_internal_gene ^ p.string("HA") ^ p.string("NA")
-p_constellation = p.regex("[TPVHX-]{6}|MIXED")
+p_constellation = p.regex("[A-Z-]{6}|MIXED|mixed")
 p_segment_number = p.regex("[1-8]")
 p_segment_subtype = p_segment ^ p_HA ^ p_NA
 

diff --git a/octofludb/formatting.py b/octofludb/formatting.py
@@ -95,6 +95,9 @@ def _make_constellations(rows: List[Tuple[str, str, str]]) -> List[Tuple[str, st
             # add flexible matching
             for c_lookup, c_letter in clade_lookup.items():
                 if c_lookup.lower() in clade.lower():
+                    log(
+                        f"{bad('WARNING:')} closest match found is partial and/or case-insensitive, {str((strain, clade, c_lookup))}"
+                    )
                     char = c_letter
                     break
             else: