WIP: Reduce lgx the alleles from the rel_dna_ser before loading to database nmdp-bioinformatics#252

Reduce "lgx" the alleles from the rel_dna_ser before loading and reduce "lgx" on input which will simplify the query and reduce the size of the table.
Failing on G group reduction of Serology. See
 # tests/features/serology.feature:15
 ```
    Given the serology typing is A10
    When reducing on the G level (ambiguous)
```
Doesn't match when serology_mapping was without lgx reduced alleles.

Author: pbashyal-nmdp

pyard/ard.py

@@ -114,7 +114,9 @@ def __init__(
         broad_splits.broad_splits_ser_mapping = (
             dr.generate_serology_broad_split_mapping(self.db_connection, imgt_version)
         )
-        dr.generate_serology_mapping(self.db_connection, imgt_version)
+        dr.generate_serology_mapping(
+            self.db_connection, imgt_version, self._redux_allele
+        )
         # Load V2 to V3 mappings
         dr.generate_v2_to_v3_mapping(self.db_connection, imgt_version)
         # Save IMGT database version
@@ -124,9 +126,6 @@ def __init__(
         # Load CIWD mapping
         dr.generate_cwd_mapping(self.db_connection)
 
-        # Close the current read-write db connection
-        self.db_connection.close()
-
         # Adjust the cache for redux
         if max_cache_size != DEFAULT_CACHE_SIZE:
             self._redux_allele = functools.lru_cache(maxsize=max_cache_size)(
@@ -147,6 +146,9 @@ def __init__(
 
             gc.freeze()
 
+        # Close the current read-write db connection
+        self.db_connection.close()
+
         # Re-open the connection in read-only mode as we're not updating it anymore
         self.db_connection, _ = db.create_db_connection(data_dir, imgt_version, ro=True)
 
@@ -268,17 +270,19 @@ def _redux_allele(
                 # If ambiguous, reduce to G group level
                 return self._redux_allele(allele, "lgx")
         elif redux_type == "S":
+            # reduce allele to ARD level
+            lgx_allele = self._redux_allele(allele, "lgx", re_ping)
             # find serology equivalent in serology_mapping
-            serology_mapping = db.find_serology_for_allele(self.db_connection, allele)
+            # look for lgx_allele as a wildcard in the allele list
+            serology_mapping = db.find_serology_for_allele(
+                self.db_connection, lgx_allele
+            )
+            # Verify that the actual allele is in allele list, if so keep track of
+            # mapped serology
             serology_set = set()
             for serology, allele_list in serology_mapping.items():
-                if allele in allele_list.split("/"):
+                if lgx_allele in allele_list.split("/"):
                     serology_set.add(serology)
-            if not serology_set and is_2_field_allele(allele):
-                for serology, allele_list in serology_mapping.items():
-                    allele_list_lgx = self.redux(allele_list, "lgx")
-                    if allele in allele_list_lgx.split("/"):
-                        serology_set.add(serology)
             return "/".join(
                 sorted(
                     serology_set, key=functools.cmp_to_key(self.smart_sort_comparator)

pyard/data_repository.py

@@ -354,7 +354,9 @@ def to_serological_name(locus_name: str):
     return sero_name
 
 
-def generate_serology_mapping(db_connection: sqlite3.Connection, imgt_version):
+def generate_serology_mapping(
+    db_connection: sqlite3.Connection, imgt_version, redux_function
+):
     if not db.table_exists(db_connection, "serology_mapping"):
         df_sero = load_serology_mappings(imgt_version)
 
@@ -388,9 +390,14 @@ def generate_serology_mapping(db_connection: sqlite3.Connection, imgt_version):
             to_serological_name
         )
 
+        sero_mapping_combined["lgx"] = sero_mapping_combined["Allele"].apply(
+            lambda allele: redux_function(allele, "lgx")
+        )
+
+        # Create a serology to lgx reduced allele list mapping
         sero_mapping = (
             sero_mapping_combined.groupby("Sero")
-            .apply(lambda x: "/".join(sorted(x["Allele"])))
+            .apply(lambda x: "/".join(set(x["lgx"])))
             .to_dict()
         )