Support DRBX Blending

 - `blender.py` blends drb1 and drbx
 - Gherkin feature files as tests
 - Fixes in Makefile and cleanup

Author: pbashyal-nmdp

Makefile

@@ -1,3 +1,6 @@
+PROJECT_NAME := $(shell basename `pwd`)
+PACKAGE_NAME := pyard
+
 .PHONY: clean clean-test clean-pyc clean-build docs help
 .DEFAULT_GOAL := help
 define BROWSER_PYSCRIPT

README.md

@@ -121,6 +121,14 @@ ard.redux_gl('B14', 'lg')
 | `exon`         | Reduce/Expand to exon level                     |
 | `U2`           | Reduce to 2 field unambiguous level             |
 
+## Perform DRB1 blending with DRB3, DRB4 and DRB5
+
+```python
+import pyard
+
+pyard.dr_blender(drb1='HLA-DRB1*03:01+DRB1*04:01', drb3='DRB3*01:01', drb4='DRB4*01:03')
+# >>> 'DRB3*01:01+DRB4*01:03'
+```
 # Command Line Tools
 
 ## Using `py-ard` from the command line

pyard/__init__.py

@@ -22,6 +22,7 @@
 #    > http://www.opensource.org/licenses/lgpl-license.php
 #
 from .pyard import ARD
+from .blender import blender as dr_blender
 
 __author__ = """NMDP Bioinformatics"""
 __version__ = "0.8.0"

pyard/blender.py

@@ -0,0 +1,225 @@
+# drbx blender
+
+
+def blender(drb1, drb3="", drb4="", drb5=""):
+    try:
+        drb1_1, drb1_2 = drb1.split("+")
+        drb1_allele_1 = drb1_1.split("*")[1]
+        drb1_allele_2 = drb1_2.split("*")[1]
+        drb1_fam_1 = drb1_allele_1.split(":")[0]
+        drb1_fam_2 = drb1_allele_2.split(":")[0]
+    except Exception:
+        return ""
+    x1 = expdrbx(drb1_fam_1)
+    x2 = expdrbx(drb1_fam_2)
+    xx = "".join(sorted([x1, x2]))
+
+    if xx == "00":
+        if drb3 != "":
+            raise DRBXBlenderError("DRB3", "none")
+        if drb4 != "":
+            raise DRBXBlenderError("DRB4", "none")
+        if drb5 != "":
+            raise DRBXBlenderError("DRB5", "none")
+        return ""
+
+    # handle 03
+    if xx == "03":
+        if drb4 != "":
+            raise DRBXBlenderError("DRB4", "none")
+        if drb5 != "":
+            raise DRBXBlenderError("DRB5", "none")
+        if drb3 != "":
+            # if 2 copies
+            drb3_g = drb3.split("+")
+            if len(drb3_g) == 2:
+                # homozygous, return one copy
+                if drb3_g[1] == drb3_g[0]:
+                    return drb3_g[0]
+                else:
+                    raise DRBXBlenderError("DRB3 het", "hom")
+            else:
+                return drb3
+        return ""
+
+    # handle 04
+    if xx == "04":
+        if drb3 != "":
+            raise DRBXBlenderError("DRB3", "none")
+        if drb5 != "":
+            raise DRBXBlenderError("DRB5", "none")
+        if drb4 != "":
+            # if 2 copies
+            drb4_g = drb4.split("+")
+            if len(drb4_g) == 2:
+                # homozygous, return one copy
+                if drb4_g[1] == drb4_g[0]:
+                    return drb4_g[0]
+                else:
+                    raise DRBXBlenderError("DRB4 het", "hom")
+            else:
+                return drb4
+        return ""
+
+    # handle 05
+    if xx == "05":
+        if drb3 != "":
+            raise DRBXBlenderError("DRB3", "none")
+        if drb4 != "":
+            raise DRBXBlenderError("DRB4", "none")
+        if drb5 != "":
+            # if 2 copies
+            drb5_g = drb5.split("+")
+            if len(drb5_g) == 2:
+                # homozygous, return one copy
+                if drb5_g[1] == drb5_g[0]:
+                    return drb5_g[0]
+                else:
+                    raise DRBXBlenderError("DRB5 het", "hom")
+            else:
+                return drb5
+        return ""
+    # handle 33
+    if xx == "33":
+        if drb4 != "":
+            raise DRBXBlenderError("DRB4", "none")
+        if drb5 != "":
+            raise DRBXBlenderError("DRB5", "none")
+        if drb3 != "":
+            return drb3
+        return ""
+
+    # handle 44
+    if xx == "44":
+        if drb3 != "":
+            raise DRBXBlenderError("DRB3", "none")
+        if drb5 != "":
+            raise DRBXBlenderError("DRB5", "none")
+        if drb4 != "":
+            return drb4
+        return ""
+
+    # handle 55
+    if xx == "55":
+        if drb3 != "":
+            raise DRBXBlenderError("DRB3", "none")
+        if drb4 != "":
+            raise DRBXBlenderError("DRB4", "none")
+        if drb5 != "":
+            return drb5
+        return ""
+
+    # handle 34
+    if xx == "34":
+        if drb5 != "":
+            raise DRBXBlenderError("DRB5", "none")
+        retg = []
+
+        if drb3 != "":
+            # if 2 copies
+            drb3_g = drb3.split("+")
+            if len(drb3_g) == 2:
+                # homozygous, return one copy
+                if drb3_g[1] == drb3_g[0]:
+                    retg.append(drb3_g[0])
+                else:
+                    raise DRBXBlenderError("DRB3 het", "hom")
+            elif len(drb3_g) == 1:
+                retg.append(drb3_g[0])
+        if drb4 != "":
+            # if 2 copies
+            drb4_g = drb4.split("+")
+            if len(drb4_g) == 2:
+                # homozygous, return one copy
+                if drb4_g[1] == drb4_g[0]:
+                    retg.append(drb4_g[0])
+                else:
+                    raise DRBXBlenderError("DRB4 het", "hom")
+            elif len(drb4_g) == 1:
+                retg.append(drb4_g[0])
+
+        return "+".join(retg)
+
+    # handle 35
+    if xx == "35":
+        if drb4 != "":
+            raise DRBXBlenderError("DRB4", "none")
+        retg = []
+
+        if drb3 != "":
+            # if 2 copies
+            drb3_g = drb3.split("+")
+            if len(drb3_g) == 2:
+                # homozygous, return one copy
+                if drb3_g[1] == drb3_g[0]:
+                    retg.append(drb3_g[0])
+                else:
+                    raise DRBXBlenderError("DRB3 het", "hom")
+            elif len(drb3_g) == 1:
+                retg.append(drb3_g[0])
+        if drb5 != "":
+            # if 2 copies
+            drb5_g = drb5.split("+")
+            if len(drb5_g) == 2:
+                # homozygous, return one copy
+                if drb5_g[1] == drb5_g[0]:
+                    retg.append(drb5_g[0])
+                else:
+                    raise DRBXBlenderError("DRB5 het", "hom")
+            elif len(drb5_g) == 1:
+                retg.append(drb5_g[0])
+
+        return "+".join(retg)
+
+    # handle 45
+    if xx == "45":
+        if drb3 != "":
+            raise DRBXBlenderError("DRB3", "none")
+        retg = []
+
+        if drb4 != "":
+            # if 2 copies
+            drb4_g = drb4.split("+")
+            if len(drb4_g) == 2:
+                # homozygous, return one copy
+                if drb4_g[1] == drb4_g[0]:
+                    retg.append(drb4_g[0])
+                else:
+                    raise DRBXBlenderError("DRB4 het", "hom")
+            elif len(drb4_g) == 1:
+                retg.append(drb4_g[0])
+        if drb5 != "":
+            # if 2 copies
+            drb5_g = drb5.split("+")
+            if len(drb5_g) == 2:
+                # homozygous, return one copy
+                if drb5_g[1] == drb5_g[0]:
+                    retg.append(drb5_g[0])
+                else:
+                    raise DRBXBlenderError("DRB5 het", "hom")
+            elif len(drb5_g) == 1:
+                retg.append(drb5_g[0])
+
+        return "+".join(retg)
+
+    print("blender fail", xx, drb1_fam_1, drb1_fam_2)
+    return ""
+
+
+def expdrbx(drb1_fam):
+    if drb1_fam in ["03", "05", "06", "11", "12", "13", "14"]:
+        return "3"
+    if drb1_fam in ["04", "07", "09"]:
+        return "4"
+    if drb1_fam in ["02", "15", "16"]:
+        return "5"
+    return "0"
+
+
+class DRBXBlenderError(Exception):
+    def __init__(self, found, expected):
+        self.found = found
+        self.expected = expected
+
+    def __str__(self):
+        return f"{self.found} where {self.expected} expected"

pyard/db.py

@@ -315,7 +315,7 @@ def similar_alleles(connection: sqlite3.Connection, allele_name: str) -> Set[str
     :param allele_name: Allele name to use as a prefix to find similar alleles
     :return: list of similar alleles
     """
-    query = f"SELECT allele FROM alleles WHERE allele LIKE ?"
+    query = "SELECT allele FROM alleles WHERE allele LIKE ?"
     cursor = connection.execute(query, (f"{allele_name}%",))
     result = cursor.fetchall()
     # fetchall() returns a list of tuples of results

pyard/misc.py

@@ -59,7 +59,3 @@ def get_P_name(a: str) -> str:
     if last_char in PandG_chars + expression_chars:
         a = a[:-1]
     return ":".join(a.split(":")[0:2]) + "P"
-
-
-def number_of_fields(allele: str) -> int:
-    return len(allele.split(":"))

setup.cfg

@@ -19,4 +19,10 @@ replace = version: "{new_version}"
 universal = 1
 
 [flake8]
-exclude = docs
+exclude =
+    venv,
+    .git,
+    __pycache__,
+    build,
+    dist,
+    docs

tests/features/drbx_blender.feature

@@ -0,0 +1,56 @@
+Feature: DRB1 blends with DRB3, DRB4, DRB5
+
+  Scenario Outline: DRB1 blends
+
+    Given a subject has <DRB1_SLUG> DRB1 SLUG
+    Given a subject has <DRB3> DRB3 allele
+    And a subject has <DRB4> DRB4 allele
+    And a subject has <DRB5> DRB5 allele
+    When I blend the DRBX alleles with DRB1 allele
+    Then it should blend as <DRBX_BLEND>
+
+    Examples: All blends with DRB1
+      | DRB1_SLUG                 | DRB3       | DRB4       | DRB5       | DRBX_BLEND            |
+      | HLA-DRB1*03:01+DRB1*04:01 | DRB3*01:01 | DRB4*01:03 | no         | DRB3*01:01+DRB4*01:03 |
+      | HLA-DRB1*03:01+DRB1*04:01 | DRB3*01:01 | DRB4*01:03 | no         | DRB3*01:01+DRB4*01:03 |
+      | HLA-DRB1*03:01+DRB1*04:01 | no         | DRB4*01:03 | no         | DRB4*01:03            |
+      | HLA-DRB1*03:01+DRB1*04:01 | DRB3*01:03 | no         | no         | DRB3*01:03            |
+      | HLA-DRB1*01:01+DRB1*08:01 | no         | no         | no         | nothing               |
+      | HLA-DRB1*01:01+DRB1*03:01 | no         | no         | no         | nothing               |
+      | HLA-DRB1*01:01+DRB1*04:01 | no         | no         | no         | nothing               |
+      | HLA-DRB1*03:01+DRB1*13:01 | DRB3*01:01 | no         | no         | DRB3*01:01            |
+      | HLA-DRB1*15:01+DRB1*16:01 | no         | no         | DRB5*01:03 | DRB5*01:03            |
+
+  Scenario Outline: DRB1 doesn't blend
+
+    Given a subject has <DRB1_SLUG> DRB1 SLUG
+    Given a subject has <DRB3> DRB3 allele
+    And a subject has <DRB4> DRB4 allele
+    And a subject has <DRB5> DRB5 allele
+    When I blend the DRBX alleles with DRB1 allele, it shouldn't blend
+    Then <Expected> was expected, but found <Found>
+
+    Examples: Doesn't blend with DRB1
+      | DRB1_SLUG                 | DRB3                  | DRB4                  | DRB5       | Expected | Found    |
+      | HLA-DRB1*03:01+DRB1*04:01 | DRB3*01:01            | DRB4*01:03            | DRB5*01:05 | none     | DRB5     |
+      | HLA-DRB1*03:01+DRB1*04:01 | DRB3*01:01+DRB3*02:01 | DRB4*01:03            | no         | hom      | DRB3 het |
+      | HLA-DRB1*03:01+DRB1*04:01 | DRB3*01:01            | DRB4*01:03+DRB4*01:05 | no         | hom      | DRB4 het |
+      | HLA-DRB1*01:01+DRB1*08:01 | DRB3*01:01            | no                    | no         | none     | DRB3     |
+      | HLA-DRB1*01:01+DRB1*08:01 | no                    | DRB4*01:01            | no         | none     | DRB4     |
+      | HLA-DRB1*01:01+DRB1*08:01 | no                    | no                    | DRB5*01:01 | none     | DRB5     |
+      | HLA-DRB1*01:01+DRB1*03:01 | DRB3*01:01            | DRB4*01:03            | no         | none     | DRB4     |
+      | HLA-DRB1*01:01+DRB1*03:01 | DRB3*01:01            | no                    | DRB5*01:03 | none     | DRB5     |
+      | HLA-DRB1*01:01+DRB1*04:01 | no                    | DRB4*01:01+DRB4*01:03 | no         | hom      | DRB4 het |
+      | HLA-DRB1*01:01+DRB1*04:01 | DRB3*01:01            | DRB4*01:03            | no         | none     | DRB3     |
+      | HLA-DRB1*01:01+DRB1*04:01 | no                    | DRB4*01:01            | DRB5*01:03 | none     | DRB5     |
+      | HLA-DRB1*03:01+DRB1*13:01 | DRB3*01:01            | DRB4*01:03            | no         | none     | DRB4     |
+      | HLA-DRB1*03:01+DRB1*13:01 | DRB3*01:01            | no                    | DRB5*01:03 | none     | DRB5     |
+      | HLA-DRB1*03:01+DRB1*13:01 | no                    | DRB4*01:01            | DRB5*01:03 | none     | DRB4     |
+      | HLA-DRB1*04:01+DRB1*09:01 | DRB3*01:01            | no                    | no         | none     | DRB3     |
+      | HLA-DRB1*04:01+DRB1*09:01 | DRB3*01:01            | DRB4*01:03            | no         | none     | DRB3     |
+      | HLA-DRB1*04:01+DRB1*09:01 | DRB3*01:01            | no                    | DRB5*01:03 | none     | DRB3     |
+      | HLA-DRB1*15:01+DRB1*16:01 | no                    | DRB4*01:01            | DRB5*01:03 | none     | DRB4     |
+      | HLA-DRB1*15:01+DRB1*16:01 | DRB3*01:01            | no                    | no         | none     | DRB3     |
+      | HLA-DRB1*15:01+DRB1*16:01 | DRB3*01:01            | DRB4*01:03            | no         | none     | DRB3     |
+      | HLA-DRB1*15:01+DRB1*16:01 | DRB3*01:01            | no                    | DRB5*01:03 | none     | DRB3     |
+      | HLA-DRB1*15:01+DRB1*16:01 | no                    | DRB4*01:01            | DRB5*01:03 | none     | DRB4     |

tests/steps/drbx_blender.py

@@ -0,0 +1,53 @@
+from behave import *
+from hamcrest import assert_that, is_, calling, raises
+
+import pyard
+from pyard.blender import DRBXBlenderError
+
+
+@given("a subject has {DRB1_SLUG} DRB1 SLUG")
+def step_impl(context, DRB1_SLUG):
+    context.drb1_slug = DRB1_SLUG
+
+
+@given("a subject has {DRB3} DRB3 allele")
+def step_impl(context, DRB3):
+    context.drb3 = DRB3 if DRB3 != "no" else ""
+
+
+@step("a subject has {DRB4} DRB4 allele")
+def step_impl(context, DRB4):
+    context.drb4 = DRB4 if DRB4 != "no" else ""
+
+
+@step("a subject has {DRB5} DRB5 allele")
+def step_impl(context, DRB5):
+    context.drb5 = DRB5 if DRB5 != "no" else ""
+
+
+@when("I blend the DRBX alleles with DRB1 allele")
+def step_impl(context):
+    context.blended_drbx = pyard.dr_blender(
+        context.drb1_slug, context.drb3, context.drb4, context.drb5
+    )
+
+
+@then("it should blend as {DRBX_BLEND}")
+def step_impl(context, DRBX_BLEND):
+    DRBX_BLEND = DRBX_BLEND if DRBX_BLEND != "nothing" else ""
+    assert_that(context.blended_drbx, is_(DRBX_BLEND))
+
+
+@when("I blend the DRBX alleles with DRB1 allele, it shouldn't blend")
+def step_impl(context):
+    pass
+
+
+@then("{expected} was expected, but found {found}")
+def step_impl(context, expected, found):
+    assert_that(
+        calling(pyard.dr_blender).with_args(
+            context.drb1_slug, context.drb3, context.drb4, context.drb5
+        ),
+        raises(DRBXBlenderError, f"{found} where {expected} expected"),
+    )