From a469cde156f59221bca9c21385da2878e7f12b05 Mon Sep 17 00:00:00 2001
From: Daniel Himmelstein <dhimmel@related.vc>
Date: Thu, 14 Oct 2021 13:39:21 -0700
Subject: [PATCH] set enable_mhc to False for rat

since we do not yet know the actual MHC/xMHC boundaries
---
 ensembl_genes/species.py | 12 +++++++++---
 tests/species_test.py    |  8 +++++++-
 2 files changed, 16 insertions(+), 4 deletions(-)

diff --git a/ensembl_genes/species.py b/ensembl_genes/species.py
index 38c07ec..d69fcba 100644
--- a/ensembl_genes/species.py
+++ b/ensembl_genes/species.py
@@ -1,5 +1,5 @@
 from dataclasses import dataclass
-from typing import Union
+from typing import Optional, Union
 
 from ensembl_genes.models import GeneForMHC
 
@@ -10,6 +10,7 @@ class Species:
     common_name: str
     reference_genome: str
     ensembl_gene_pattern: str
+    enable_mhc: bool
     mhc_chromosome: str
     mhc_lower: int
     mhc_upper: int
@@ -17,8 +18,10 @@ class Species:
     xmhc_upper: int
     chromosomes: list[str]
 
-    def get_mhc_category(self, gene: GeneForMHC) -> str:
+    def get_mhc_category(self, gene: GeneForMHC) -> Optional[str]:
         """Assign MHC status of MHC, xMHC, or no to an ensembl gene record."""
+        if not self.enable_mhc:
+            return None
         import pandas as pd
 
         chromosome: str = gene.chromosome
@@ -48,6 +51,7 @@ def get_mhc_category(self, gene: GeneForMHC) -> str:
     ensembl_gene_pattern=r"^ENSG[0-9]{11}$",
     # Refs MHC boundary discussion internal Related Sciences issue 127.
     # https://bioinformatics.stackexchange.com/a/14719/9750
+    enable_mhc=True,
     mhc_chromosome="6",
     mhc_lower=28_510_120,
     mhc_upper=33_480_577,
@@ -65,7 +69,9 @@ def get_mhc_category(self, gene: GeneForMHC) -> str:
     # https://github.com/related-sciences/ensembl-genes/issues/4#issuecomment-941556912
     ensembl_gene_pattern=r"^ENSRNOG[0-9]{11}$",
     # FIXME: mhc coordinates
-    mhc_chromosome="6",
+    # https://github.com/related-sciences/ensembl-genes/pull/6#discussion_r729259953
+    enable_mhc=False,
+    mhc_chromosome="20",
     mhc_lower=28_510_120,
     mhc_upper=33_480_577,
     xmhc_lower=25_726_063,
diff --git a/tests/species_test.py b/tests/species_test.py
index dd789d9..02daf88 100644
--- a/tests/species_test.py
+++ b/tests/species_test.py
@@ -1,7 +1,7 @@
 import pytest
 
 from ensembl_genes.models import GeneForMHC as Gene
-from ensembl_genes.species import get_species, human
+from ensembl_genes.species import get_species, human, rat
 
 
 def test_get_species() -> None:
@@ -31,3 +31,9 @@ def test_get_species() -> None:
 )
 def test_get_mhc_category_human(gene: Gene, category: str) -> None:
     assert human.get_mhc_category(gene) == category
+
+
+def test_get_mhc_category_rat() -> None:
+    gene = Gene("20", 0, 0)
+    assert rat.enable_mhc is False
+    assert rat.get_mhc_category(gene) is None