Consolidate alleles and code mappings into their own.

Author: pbashyal-nmdp

pyard/ard.py

@@ -96,17 +96,16 @@ def __init__(
         self.ars_mappings = dr.generate_ars_mapping(self.db_connection, imgt_version)
         # Load Alleles and XX Codes
         (
-            self.valid_alleles,
-            self.who_alleles,
-            self.xx_codes,
-            self.who_group,
-            self.exp_alleles,
+            self.code_mappings,
+            self.allele_group,
         ) = dr.generate_alleles_and_xx_codes_and_who(
             self.db_connection, imgt_version, self.ars_mappings
         )
 
         # Generate short nulls from WHO mapping
-        self.shortnulls = dr.generate_short_nulls(self.db_connection, self.who_group)
+        self.shortnulls = dr.generate_short_nulls(
+            self.db_connection, self.code_mappings.who_group
+        )
 
         # Load Serology mappings
         broad_splits.broad_splits_ser_mapping = (
@@ -214,8 +213,10 @@ def _redux_allele(
             # new redux_type which is full WHO expansion
             if self._is_who_allele(allele):
                 return allele
-            if allele in self.who_group:
-                return self.redux("/".join(self.who_group[allele]), redux_type)
+            if allele in self.code_mappings.who_group:
+                return self.redux(
+                    "/".join(self.code_mappings.who_group[allele]), redux_type
+                )
             else:
                 return allele
         elif redux_type == "exon":
@@ -354,11 +355,13 @@ def redux(self, glstring: str, redux_type: VALID_REDUCTION_TYPES) -> str:
             if self.is_XX(glstring, loc_antigen, code):
                 if is_hla_prefix:
                     reduced_alleles = self.redux(
-                        "/".join(self.xx_codes[loc_antigen]), redux_type
+                        "/".join(self.code_mappings.xx_codes[loc_antigen]), redux_type
                     )
                     return "/".join(["HLA-" + a for a in reduced_alleles.split("/")])
                 else:
-                    return self.redux("/".join(self.xx_codes[loc_antigen]), redux_type)
+                    return self.redux(
+                        "/".join(self.code_mappings.xx_codes[loc_antigen]), redux_type
+                    )
 
         # Handle MAC
         if self._config["reduce_MAC"] and self.is_mac(glstring):
@@ -403,7 +406,7 @@ def is_XX(self, glstring: str, loc_antigen: str = None, code: str = None) -> boo
                 loc_antigen, code = loc_allele[0], loc_allele[1]
             else:
                 return False
-        return code == "XX" and loc_antigen in self.xx_codes
+        return code == "XX" and loc_antigen in self.code_mappings.xx_codes
 
     def is_serology(self, allele: str) -> bool:
         """
@@ -469,15 +472,15 @@ def _is_who_allele(self, allele):
         :param allele: Allele to test
         :return: bool to indicate if allele is valid
         """
-        return allele in self.who_alleles
+        return allele in self.allele_group.who_alleles
 
     def _is_valid_allele(self, allele):
         """
         Test if allele is valid in the current imgt database
         :param allele: Allele to test
         :return: bool to indicate if allele is valid
         """
-        return allele in self.valid_alleles
+        return allele in self.allele_group.alleles
 
     def is_shortnull(self, allele):
         """
@@ -494,7 +497,7 @@ def is_exp_allele(self, allele):
         :param allele: Allele to test
         :return: bool to indicate if allele is valid
         """
-        return allele in self.exp_alleles
+        return allele in self.allele_group.exp_alleles
 
     def _get_alleles(self, code, locus_antigen) -> Iterable[str]:
         """

pyard/data_repository.py

@@ -35,7 +35,14 @@
     load_latest_version,
 )
 from .constants import expression_chars
-from .mappings import ars_mapping_tables, ARSMapping, code_mapping_tables
+from .mappings import (
+    ars_mapping_tables,
+    ARSMapping,
+    code_mapping_tables,
+    AlleleGroups,
+    CodeMappings,
+    allele_tables,
+)
 from .misc import (
     get_2field_allele,
     get_3field_allele,
@@ -63,7 +70,7 @@ def expression_reduce(df):
     return None
 
 
-def generate_ars_mapping(db_connection: sqlite3.Connection, imgt_version):
+def generate_ars_mapping(db_connection: sqlite3.Connection, imgt_version) -> ARSMapping:
     if db.tables_exist(db_connection, ars_mapping_tables):
         return db.load_ars_mappings(db_connection)
 
@@ -162,7 +169,7 @@ def generate_ars_mapping(db_connection: sqlite3.Connection, imgt_version):
 def generate_alleles_and_xx_codes_and_who(
     db_connection: sqlite3.Connection, imgt_version, ars_mappings
 ):
-    if db.tables_exist(db_connection, code_mapping_tables):
+    if db.tables_exist(db_connection, code_mapping_tables + allele_tables):
         return db.load_code_mappings(db_connection)
 
     import pandas as pd
@@ -256,7 +263,14 @@ def generate_alleles_and_xx_codes_and_who(
         who_alleles,
     )
 
-    return valid_alleles, who_alleles, xx_codes, who_group, exp_alleles
+    return (
+        CodeMappings(xx_codes=xx_codes, who_group=who_group),
+        AlleleGroups(
+            alleles=valid_alleles, who_alleles=who_alleles, exp_alleles=exp_alleles
+        ),
+    )
+
+    # return valid_alleles, who_alleles, xx_codes, who_group, exp_alleles
 
 
 def generate_short_nulls(db_connection, who_group):

pyard/db.py

@@ -24,7 +24,7 @@
 import sqlite3
 from typing import Tuple, Dict, Set, List
 
-from .mappings import ARSMapping
+from .mappings import ARSMapping, CodeMappings, AlleleGroups
 from .misc import get_imgt_db_versions, get_default_db_directory
 
 
@@ -467,7 +467,12 @@ def load_code_mappings(db_connection):
     xx_codes = {k: v.split("/") for k, v in xx_codes.items()}
     exp_alleles = load_dict(db_connection, "exp_alleles", ("exp_allele", "allele_list"))
     exp_alleles = {k: v.split("/") for k, v in exp_alleles.items()}
-    return valid_alleles, who_alleles, xx_codes, who_group, exp_alleles
+    return (
+        CodeMappings(xx_codes=xx_codes, who_group=who_group),
+        AlleleGroups(
+            alleles=valid_alleles, who_alleles=who_alleles, exp_alleles=exp_alleles
+        ),
+    )
 
 
 def load_shortnulls(db_connection):

pyard/mappings.py

@@ -9,12 +9,18 @@
     "exon_group",
     "p_not_g",
 ]
+
 code_mapping_tables = [
+    "xx_codes",
+    "who_group",
+]
+
+allele_tables = [
     "alleles",
     "exp_alleles",
-    "xx_codes",
     "who_alleles",
-    "who_group",
 ]
 
 ARSMapping = namedtuple("ARSMapping", ars_mapping_tables)
+CodeMappings = namedtuple("CodeMappings", code_mapping_tables)
+AlleleGroups = namedtuple("AlleleGroups", allele_tables)

scripts/pyard-status

@@ -96,6 +96,7 @@ if __name__ == "__main__":
             for table in (
                 pyard.mappings.ars_mapping_tables
                 + pyard.mappings.code_mapping_tables
+                + pyard.mappings.allele_tables
                 + ["mac_codes"]
             ):
                 if db.table_exists(db_connection, table):