From bae82d20cce34045c32936972ba68a385b6f8209 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?F=C3=A9lix=20Raimundo?= <gamaz3ps@gmail.com>
Date: Thu, 4 Apr 2024 11:35:08 -0400
Subject: [PATCH] Some cleanup on the ensembldb code (#67)

* Some cleanup on the ensembldb code

* fix __repr__

* fix comments

* Update src/genomic_features/ensembl/ensembldb.py

---------

Co-authored-by: Felix Raimundo <felix.raimundo@3umassmed.edu>
Co-authored-by: Isaac Virshup <ivirshup@gmail.com>
---
 src/genomic_features/ensembl/ensembldb.py | 19 +++++++++++--------
 tests/test_basic.py                       |  2 +-
 2 files changed, 12 insertions(+), 9 deletions(-)

diff --git a/src/genomic_features/ensembl/ensembldb.py b/src/genomic_features/ensembl/ensembldb.py
index 5d96d29..93bc8ae 100644
--- a/src/genomic_features/ensembl/ensembldb.py
+++ b/src/genomic_features/ensembl/ensembldb.py
@@ -20,10 +20,7 @@
 PKG_CACHE_DIR = "genomic-features"
 
 BIOC_ANNOTATION_HUB_URL = (
-    "https://bioconductorhubs.blob.core.windows.net/annotationhub/"
-)
-ENSEMBL_URL_TEMPLATE = (
-    BIOC_ANNOTATION_HUB_URL + "AHEnsDbs/v{version}/EnsDb.{species}.v{version}.sqlite"
+    "https://bioconductorhubs.blob.core.windows.net/annotationhub"
 )
 ANNOTATION_HUB_URL = (
     "https://annotationhub.bioconductor.org/metadata/annotationhub.sqlite3"
@@ -56,7 +53,7 @@ def annotation(
     """
     try:
         sqlite_file_path = retrieve_annotation(
-            ENSEMBL_URL_TEMPLATE.format(species=species, version=version)
+            f'{BIOC_ANNOTATION_HUB_URL}/AHEnsDbs/v{version}/EnsDb.{species}.v{version}.sqlite'
         )
 
         if backend == "sqlite":
@@ -74,7 +71,8 @@ def annotation(
     except HTTPError as err:
         if err.response.status_code == 404:
             raise ValueError(
-                f"No Ensembl database found for {species} v{version}. Check available versions with `genomic_features.ensembl.list_versions`."
+                f"No Ensembl database found for {species} v{version}. Check "
+                f"available versions with `genomic_features.ensembl.list_ensdb_annotations `."
             ) from err
         else:
             raise HTTPError from err
@@ -125,7 +123,8 @@ def list_ensdb_annotations(species: None | str | list[str] = None) -> DataFrame:
         # check that species exist
         if version_table.shape[0] == 0:
             raise ValueError(
-                f"No Ensembl database found for {species}. Check species name."
+                f"No Ensembl database found for {species}. Available species can "
+                f"be found via: `list_ensdb_annotations()['Species'].unique()`."
             )
 
     version_table["Ensembl_version"] = version_table["rdatapath"].str.split(
@@ -153,7 +152,11 @@ def metadata(self) -> dict:
 
     def __repr__(self) -> str:
         d = self.metadata
-        return f"EnsemblDB(organism='{d['Organism']}', ensembl_release='{d['ensembl_version']}')"
+        return (
+            f"EnsemblDB(organism='{d['Organism']}', "
+            f"ensembl_release='{d['ensembl_version']}', "
+            f"genome_build='{d['genome_build']}')"
+        )
 
     def genes(
         self,
diff --git a/tests/test_basic.py b/tests/test_basic.py
index 3e90722..c127400 100644
--- a/tests/test_basic.py
+++ b/tests/test_basic.py
@@ -20,7 +20,7 @@ def test_missing_version():
 
 def test_repr():
     result = repr(gf.ensembl.annotation("Hsapiens", 108))
-    expected = "EnsemblDB(organism='Homo sapiens', ensembl_release='108')"
+    expected = "EnsemblDB(organism='Homo sapiens', ensembl_release='108', genome_build='GRCh38')"
 
     assert result == expected