add support for rat species

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
 
@@ -10,15 +10,18 @@ class Species:
     common_name: str
     reference_genome: str
     ensembl_gene_pattern: str
+    enable_mhc: bool
     mhc_chromosome: str
     mhc_lower: int
     mhc_upper: int
     xmhc_lower: int
     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
@@ -48,17 +51,38 @@ def get_mhc_category(self, gene: GeneForMHC) -> str:
     ensembl_gene_pattern=r"^ENSG[0-9]{11}$",
     # Refs MHC boundary discussion internal Related Sciences issue 127.
     # https://bioinformatics.stackexchange.com/a/14719/9750
+    enable_mhc=True,
     mhc_chromosome="6",
     mhc_lower=28_510_120,
     mhc_upper=33_480_577,
     xmhc_lower=25_726_063,
     xmhc_upper=33_410_226,
     # Chromosome names applied to genes on the primary assembly rather than alternative sequences.
     # Refs internal Related Sciences issue 241.
-    chromosomes=[*map(str, range(1, 23)), "X", "Y", "MT"],
+    chromosomes=[*map(str, range(1, 22 + 1)), "X", "Y", "MT"],
 )
+rat = Species(
+    name="rattus_norvegicus",
+    common_name="rat",
+    # Rnor_6.0
+    reference_genome="6",
+    # https://github.com/related-sciences/ensembl-genes/issues/4#issuecomment-941556912
+    ensembl_gene_pattern=r"^ENSRNOG[0-9]{11}$",
+    # FIXME: mhc coordinates
+    # 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,
+    xmhc_upper=33_410_226,
+    # Chromosome names applied to genes on the primary assembly rather than alternative sequences.
+    # Refs internal Related Sciences issue 241.
+    chromosomes=[*map(str, range(1, 20 + 1)), "X", "Y", "MT"],
+)
+
 
-all_species = [human]
+all_species = [human, rat]
 
 
 def get_species(species: Union[str, Species]) -> Species:

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