From c44714058b8977f24f9932dc5bdb4b5fd0b808fe Mon Sep 17 00:00:00 2001
From: silastittes <silas.tittes@gmail.com>
Date: Thu, 9 Jan 2025 11:43:12 -0800
Subject: [PATCH] adding assembly_name attribute

---
 stdpopsim/annotations.py                 | 4 ++++
 stdpopsim/catalog/AraTha/annotations.py  | 3 +++
 stdpopsim/catalog/DroMel/annotations.py  | 2 ++
 stdpopsim/catalog/HomSap/annotations.py  | 2 ++
 stdpopsim/catalog/HomSap/genetic_maps.py | 5 +++++
 stdpopsim/catalog/PhoSin/annotations.py  | 2 ++
 stdpopsim/genetic_maps.py                | 2 ++
 7 files changed, 20 insertions(+)

diff --git a/stdpopsim/annotations.py b/stdpopsim/annotations.py
index 54dd15de8..71dfcc7c0 100644
--- a/stdpopsim/annotations.py
+++ b/stdpopsim/annotations.py
@@ -27,6 +27,9 @@ class Annotation:
     :vartype citations: list of :class:`.Citation`
     :ivar file_pattern: The pattern used to map individual chromosome id strings
         to files
+        :ivar assembly_name: The name of the genome assembly.
+    :vartype assembly_name: str    
+    :ivar str assembly_name: The name of the assembly the annotation is based on
     """
 
     id = attr.ib()
@@ -40,6 +43,7 @@ class Annotation:
     file_pattern = attr.ib()
     annotation_source = attr.ib()
     annotation_type = attr.ib()
+    assembly_name = attr.ib()
 
     def __attrs_post_init__(self):
         self._cache = stdpopsim.CachedData(
diff --git a/stdpopsim/catalog/AraTha/annotations.py b/stdpopsim/catalog/AraTha/annotations.py
index 612fa3275..8e8e4e590 100644
--- a/stdpopsim/catalog/AraTha/annotations.py
+++ b/stdpopsim/catalog/AraTha/annotations.py
@@ -27,6 +27,7 @@
     file_pattern="araport_exons_{id}.txt",
     annotation_source="araport11",
     annotation_type="exon",
+    assembly_name = "TAIR10"
 )
 _species.add_annotations(_an)
 
@@ -56,5 +57,7 @@
     file_pattern="araport_CDS_{id}.txt",
     annotation_source="araport11",
     annotation_type="CDS",
+    assembly_name = "TAIR10"
+
 )
 _species.add_annotations(_an2)
diff --git a/stdpopsim/catalog/DroMel/annotations.py b/stdpopsim/catalog/DroMel/annotations.py
index 9f7c3abaa..b378bf7ac 100644
--- a/stdpopsim/catalog/DroMel/annotations.py
+++ b/stdpopsim/catalog/DroMel/annotations.py
@@ -28,6 +28,7 @@
     file_pattern="flybase_exons_{id}.txt",
     annotation_source="FlyBase",
     annotation_type="exon",
+    assembly_name="BDGP6.32.51",
 )
 _species.add_annotations(_an)
 
@@ -57,5 +58,6 @@
     file_pattern="flybase_CDS_{id}.txt",
     annotation_source="FlyBase",
     annotation_type="CDS",
+    assembly_name="BDGP6.32.51",
 )
 _species.add_annotations(_an2)
diff --git a/stdpopsim/catalog/HomSap/annotations.py b/stdpopsim/catalog/HomSap/annotations.py
index ace4a53cd..bd494dacf 100644
--- a/stdpopsim/catalog/HomSap/annotations.py
+++ b/stdpopsim/catalog/HomSap/annotations.py
@@ -27,6 +27,7 @@
     file_pattern="ensembl_havana_exons_{id}.txt",
     annotation_source="ensembl_havana",
     annotation_type="exon",
+    assembly_name = "GRCh38.p13"
 )
 _species.add_annotations(_an)
 
@@ -56,5 +57,6 @@
     file_pattern="ensembl_havana_CDS_{id}.txt",
     annotation_source="ensembl_havana",
     annotation_type="CDS",
+    assembly_name = "GRCh38.p13",
 )
 _species.add_annotations(_an2)
diff --git a/stdpopsim/catalog/HomSap/genetic_maps.py b/stdpopsim/catalog/HomSap/genetic_maps.py
index f21368919..300eb6686 100644
--- a/stdpopsim/catalog/HomSap/genetic_maps.py
+++ b/stdpopsim/catalog/HomSap/genetic_maps.py
@@ -31,6 +31,7 @@
     sha256="80f22d9e6cb0e497074ed1bc277e765fa9d8e22f21b2f66c3b10286520f6b68f",
     file_pattern="genetic_map_GRCh37_chr{id}.txt",
     citations=[_hapmap2007.because(stdpopsim.CiteReason.GEN_MAP)],
+    assembly_name="GRCh37",
 )
 _species.add_genetic_map(_gm)
 
@@ -60,6 +61,7 @@
     sha256="497512ed1c0f8a40e9aa13696049a9f8c3cb062e898921cfd7d85ce9d14c4baa",
     file_pattern="genetic_map_Hg38_chr{id}.txt",
     citations=[_hapmap2007.because(stdpopsim.CiteReason.GEN_MAP)],
+    assembly_name="GRCh38.p13",
 )
 _species.add_genetic_map(_gm)
 
@@ -90,6 +92,7 @@
             reasons={stdpopsim.CiteReason.GEN_MAP},
         )
     ],
+    assembly_name="GRCh36",
 )
 _species.add_genetic_map(_gm)
 
@@ -131,6 +134,7 @@
             reasons={stdpopsim.CiteReason.GEN_MAP},
         )
     ],
+    assembly_name="GRCh38.p13",
 )
 _species.add_genetic_map(_gm)
 
@@ -223,5 +227,6 @@
                 reasons={stdpopsim.CiteReason.GEN_MAP},
             )
         ],
+        assembly_name="GRCh38.p13",
     )
     _species.add_genetic_map(_gm)
diff --git a/stdpopsim/catalog/PhoSin/annotations.py b/stdpopsim/catalog/PhoSin/annotations.py
index 55f2c1a9a..07dadf31a 100644
--- a/stdpopsim/catalog/PhoSin/annotations.py
+++ b/stdpopsim/catalog/PhoSin/annotations.py
@@ -27,6 +27,7 @@
     file_pattern="Phocoena_sinus.mPhoSin1.pri.110_exon_{id}.txt",
     annotation_source="ensembl",
     annotation_type="exon",
+    assembly_name="mPhoSin1.pri",
 )
 _species.add_annotations(_an)
 
@@ -56,5 +57,6 @@
     file_pattern="Phocoena_sinus.mPhoSin1.pri.110_CDS_{id}.txt",
     annotation_source="ensembl",
     annotation_type="CDS",
+    assembly_name="mPhoSin1.pri",
 )
 _species.add_annotations(_an2)
diff --git a/stdpopsim/genetic_maps.py b/stdpopsim/genetic_maps.py
index c6837d5d4..e6669d669 100644
--- a/stdpopsim/genetic_maps.py
+++ b/stdpopsim/genetic_maps.py
@@ -35,6 +35,7 @@ def __init__(
         description=None,
         long_description=None,
         citations=None,
+        assembly_name=None,
     ):
         self.id = id
         self.species = species
@@ -45,6 +46,7 @@ def __init__(
         self.file_pattern = file_pattern
         self.description = description
         self.citations = citations
+        self.assembly_name = assembly_name
 
         self._cache = stdpopsim.CachedData(
             namespace=f"genetic_maps/{self.species.id}/{id}",