Merge pull request #280 from pbashyal-nmdp/fix_mica_micb

Fix Single field MICA, MICB Alleles failing

Author: mmaiers-nmdp

pyard/ard.py

@@ -43,6 +43,7 @@
     VALID_REDUCTION_TYPES,
     expression_chars,
     DEFAULT_CACHE_SIZE,
+    G_GROUP_LOCI,
 )
 
 default_config = {
@@ -387,6 +388,10 @@ def redux(self, glstring: str, redux_type: VALID_REDUCTION_TYPES) -> str:
             loc_allele = glstring.split(":")
             loc_antigen, code = loc_allele[0], loc_allele[1]
         else:
+            if "*" in glstring:
+                locus, _ = glstring.split("*")
+                if locus not in G_GROUP_LOCI:
+                    return glstring
             raise InvalidTypingError(
                 f"{glstring} is not a valid V2 or Serology typing."
             )
@@ -534,6 +539,7 @@ def is_v2(self, allele: str) -> bool:
             self._config["reduce_v2"]
             and "*" in allele
             and ":" not in allele
+            and allele.split("*")[0] not in ["MICA", "MICB", "HFE"]
             and allele != self._map_v2_to_v3(allele)
         )
 

pyard/constants.py

@@ -25,7 +25,32 @@
 
 HLA_regex = re.compile("^HLA-")
 
-VALID_REDUCTION_TYPES = ["G", "P", "lg", "lgx", "W", "exon", "U2", "S"]
-expression_chars = ["N", "Q", "L", "S"]
+VALID_REDUCTION_TYPES = ("G", "P", "lg", "lgx", "W", "exon", "U2", "S")
+expression_chars = ("N", "Q", "L", "S")
 # List of P and G characters
-PandG_chars = ["P", "G"]
+P_and_G_chars = ("P", "G")
+
+# Loci with G group data
+# Retrieved from lgx_group
+# sqlite> select distinct(substr(allele, 1, instr(allele, '*') - 1)) from lgx_group;
+G_GROUP_LOCI = (
+    "A",
+    "B",
+    "C",
+    "DMA",
+    "DMB",
+    "DOA",
+    "DOB",
+    "DPA1",
+    "DPB1",
+    "DQA1",
+    "DQB1",
+    "DRA",
+    "DRB1",
+    "DRB3",
+    "DRB4",
+    "DRB5",
+    "E",
+    "F",
+    "G",
+)

pyard/misc.py

@@ -24,7 +24,7 @@
 import tempfile
 from typing import List
 
-from pyard.constants import VALID_REDUCTION_TYPES, expression_chars, PandG_chars
+from pyard.constants import VALID_REDUCTION_TYPES, expression_chars, P_and_G_chars
 
 
 def get_n_field_allele(allele: str, n: int, preserve_expression=False) -> str:
@@ -47,15 +47,15 @@ def get_n_field_allele(allele: str, n: int, preserve_expression=False) -> str:
 
 def get_3field_allele(a: str) -> str:
     last_char = a[-1]
-    if last_char in PandG_chars:
+    if last_char in P_and_G_chars:
         a = a[:-1]
 
     return get_n_field_allele(a, 3)
 
 
 def get_2field_allele(a: str) -> str:
     last_char = a[-1]
-    if last_char in PandG_chars:
+    if last_char in P_and_G_chars:
         a = a[:-1]
     return get_n_field_allele(a, 2)
 
@@ -76,7 +76,7 @@ def is_2_field_allele(allele: str) -> bool:
 def get_G_name(a: str) -> str:
     a = a.split("/")[0]
     last_char = a[-1]
-    if last_char in PandG_chars + expression_chars:
+    if last_char in P_and_G_chars + expression_chars:
         a = a[:-1]
     if len(a.split(":")) == 2:
         return ":".join([a, "01"]) + "G"
@@ -88,7 +88,7 @@ def get_G_name(a: str) -> str:
 def get_P_name(a: str) -> str:
     a = a.split("/")[0]
     last_char = a[-1]
-    if last_char in PandG_chars + expression_chars:
+    if last_char in P_and_G_chars + expression_chars:
         a = a[:-1]
     return ":".join(a.split(":")[0:2]) + "P"
 

tests/features/allele.feature

@@ -71,3 +71,19 @@ Feature: Alleles
       | Allele      | Level | Redux Allele |
       | A*24:329    | lgx   | A*24:329Q    |
       | DQB1*03:276 | lgx   | DQB1*03:01   |
+
+
+  Scenario Outline: Single field MICA, MICB Alleles
+
+  For MICA, MICB alleles with single field, their reduced version is self.
+
+    Given the allele as <Allele>
+    When reducing on the <Level> level
+    Then the reduced allele is found to be <Redux Allele>
+
+    Examples:
+      | Allele   | Level | Redux Allele |
+      | HFE*002  | lgx   | HFE*002      |
+      | MICA*040 | lgx   | MICA*040     |
+      | MICB*006 | lgx   | MICB*006     |
+      | MICB*029 | lgx   | MICB*029     |