Merge pull request #63 from pbashyal-nmdp/fix_serological_mapping

Fix serological mapping

Author: mmaiers-nmdp

pyard/__init__.py

@@ -24,4 +24,4 @@
 from .pyard import ARD
 
 __author__ = """NMDP Bioinformatics"""
-__version__ = '0.4.1'
+__version__ = '0.5.0'

pyard/data_repository.py

@@ -240,6 +240,23 @@ def generate_mac_codes(db_connection: sqlite3.Connection, refresh_mac: bool):
                      dictionary=mac, columns=('code', 'alleles'))
 
 
+def to_serological_name(locus_name: str):
+    """
+    Map a DNA Allele name to Serological Equivalent.
+    http://hla.alleles.org/antigens/recognised_serology.html
+    Eg:
+      A*1 -> A1
+      ...
+      DRB5*51 -> DR51
+    :param locus_name: DNA Locus Name
+    :return: Serological equivalent
+    """
+    locus, sero_number = locus_name.split('*')
+    sero_locus = locus[:2]
+    sero_name = sero_locus + sero_number
+    return sero_name
+
+
 def generate_serology_mapping(db_connection: sqlite3.Connection, imgt_version):
     if not db.table_exists(db_connection, 'serology_mapping'):
         # Load WMDA serology mapping data
@@ -270,8 +287,13 @@ def generate_serology_mapping(db_connection: sqlite3.Connection, imgt_version):
         sero_mapping_combined = pd.concat([usa[['Sero', 'Allele']],
                                            psa[['Sero', 'Allele']],
                                            asa[['Sero', 'Allele']]])
-        sero_mapping = sero_mapping_combined.groupby('Sero').\
-            apply(lambda x: '/'.join(sorted(x['Allele']))).\
+
+        # Map to only valid serological antigen name
+        sero_mapping_combined['Sero'] = sero_mapping_combined['Sero']. \
+            apply(to_serological_name)
+
+        sero_mapping = sero_mapping_combined.groupby('Sero'). \
+            apply(lambda x: '/'.join(sorted(x['Allele']))). \
             to_dict()
 
         # Save the serology mapping to db

pyard/db.py

@@ -79,7 +79,7 @@ def mac_code_to_alleles(connection: sqlite3.Connection, code: str) -> List[str]:
     if result:
         alleles = result[0].split('/')
     else:
-        alleles = None
+        alleles = []
     return alleles
 
 
@@ -98,7 +98,7 @@ def serology_to_alleles(connection: sqlite3.Connection, serology: str) -> List[s
     if result:
         alleles = result[0].split('/')
     else:
-        alleles = None
+        alleles = []
     return alleles
 
 

pyard/pyard.py

@@ -174,20 +174,14 @@ def redux_gl(self, glstring: str, redux_type: str) -> str:
 
         # Handle Serology
         if self.is_serology(glstring):
-            if HLA_regex.search(glstring):
-                # Remove HLA- prefix
-                serology = glstring.split("-")[1]
-                alleles = self._get_alleles_from_serology(serology)
-                alleles = ['HLA-' + a for a in alleles]
-            else:
-                alleles = self._get_alleles_from_serology(glstring)
+            alleles = self._get_alleles_from_serology(glstring)
             return self.redux_gl("/".join(alleles), redux_type)
 
         loc_allele = glstring.split(":")
         loc_name, code = loc_allele[0], loc_allele[1]
 
         # Handle XX codes
-        if (self.is_mac(glstring) and glstring.split(":")[1] == "XX") and loc_name in self.xx_codes:
+        if self.is_mac(glstring) and code == "XX" and loc_name in self.xx_codes:
             return self.redux_gl("/".join(self.xx_codes[loc_name]), redux_type)
 
         # Handle MAC
@@ -207,11 +201,26 @@ def redux_gl(self, glstring: str, redux_type: str) -> str:
     @staticmethod
     def is_serology(allele: str) -> bool:
         """
-        An allele is serology if the allele name after * is numeral only, no ':'
+        A serology has the locus name (first 2 letters for DRB1, DRB3, DQB1, DQA1, DPB1 and DPA1)
+        of the allele followed by numerical antigen.
+
         :param allele: The allele to test for serology
         :return: True if serology
         """
-        return allele.split('*')[1].isdigit()
+        if '*' in allele or ':' in allele:
+            return False
+
+        locus = allele[0:2]
+        if locus in ['DR', 'DP', 'DQ']:
+            antigen = allele[2:]
+            return antigen.isdigit()
+
+        locus = allele[0:1]
+        if locus in ['A', 'B', 'C', 'D']:
+            antigen = allele[1:]
+            return antigen.isdigit()
+
+        return False
 
     @staticmethod
     def is_mac(gl: str) -> bool:
@@ -229,7 +238,9 @@ def _is_valid_allele(self, allele):
         :param allele: Allele to test
         :return: bool to indicate if allele is valid
         """
-        return allele in self.valid_alleles
+        if self._remove_invalid:
+            return allele in self.valid_alleles
+        return True
 
     def _get_alleles(self, code, loc_name) -> Iterable[str]:
         """
@@ -239,12 +250,18 @@ def _get_alleles(self, code, loc_name) -> Iterable[str]:
         :return: valid alleles corresponding to allele code
         """
         alleles = mac_code_to_alleles(self.db_connection, code)
-        return filter(self._is_valid_allele,
-                      [f'{loc_name}:{a}' for a in alleles])
+        if self._remove_invalid:
+            return filter(self._is_valid_allele,
+                          [f'{loc_name}:{a}' for a in alleles])
+        else:
+            return [f'{loc_name}:{a}' for a in alleles]
 
     def _get_alleles_from_serology(self, serology) -> Iterable[str]:
         alleles = db.serology_to_alleles(self.db_connection, serology)
-        return filter(self._is_valid_allele, alleles)
+        if self._remove_invalid:
+            return filter(self._is_valid_allele, alleles)
+        else:
+            return alleles
 
     def isvalid(self, allele: str) -> bool:
         """
@@ -255,6 +272,8 @@ def isvalid(self, allele: str) -> bool:
         :return: allele or empty
         :rtype: bool
         """
+        if allele == '':
+            return False
         if not self.is_mac(allele) and not self.is_serology(allele):
             # Alleles ending with P or G are valid_alleles
             if allele.endswith(('P', 'G')):
@@ -310,7 +329,6 @@ def mac_toG(self, allele: str) -> str:
                 return ''
             else:
                 return "/".join(group)
-
         else:
             return ''
 

setup.cfg

@@ -1,5 +1,5 @@
 [bumpversion]
-current_version = 0.4.1
+current_version = 0.5.0
 commit = True
 tag = True
 

setup.py

@@ -42,7 +42,7 @@
 
 setup(
     name='py-ard',
-    version='0.4.1',
+    version='0.5.0',
     description="ARD reduction for HLA with python",
     long_description=readme + '\n\n' + history,
     author="CIBMTR",

tests/features/serology.feature

@@ -12,13 +12,9 @@ Feature: Serology
 
     Examples: Valid A serology typings
       | Serology | Level | Redux Allele                                                |
-      | A*10     | G     | A*26:01:01G/A*26:10/A*26:15/A*26:92/A*66:01:01G/A*66:03:01G |
-      | A*10     | lg    | A*26:01g/A*26:10g/A*26:15g/A*26:92g/A*66:01g/A*66:03g       |
-      | A*10     | lgx   | A*26:01/A*26:10/A*26:15/A*26:92/A*66:01/A*66:03             |
-
-    Examples: With HLA- prefix
-      | Serology    | Level | Redux Allele                                                                        |
-      | HLA-A*10    | G     | HLA-A*26:01:01G/HLA-A*26:10/HLA-A*26:15/HLA-A*26:92/HLA-A*66:01:01G/HLA-A*66:03:01G |
-      | HLA-B*15:03 | G     | HLA-B*15:03:01G                                                                     |
-      | HLA-DQB1*1  | G     | HLA-DQB1*06:11:01/HLA-DQB1*06:11:02/HLA-DQB1*06:11:03/HLA-DQB1*06:12                |
-      | HLA-DQB1*1  | lg    | HLA-DQB1*06:11g/HLA-DQB1*06:12g                                                     |
+      | A10      | G     | A*26:01:01G/A*26:10/A*26:15/A*26:92/A*66:01:01G/A*66:03:01G |
+      | A10      | lg    | A*26:01g/A*26:10g/A*26:15g/A*26:92g/A*66:01g/A*66:03g       |
+      | A10      | lgx   | A*26:01/A*26:10/A*26:15/A*26:92/A*66:01/A*66:03             |
+      | A19      | G     | A*02:65/A*33:09                                             |
+      | DR1403   | G     | DRB1*14:03:01/DRB1*14:03:02                                 |
+      | DR2      | G     | DRB1*15:08/DRB1*16:03                                       |