update unit test and add build script

.github/workflows/test.yaml

@@ -20,11 +20,11 @@ jobs:
           pip install .
           cd test
           ../bin/agfusion download -g hg19
-          ../bin/agfusion download -s homo_sapiens -r 95
+          ../bin/agfusion download -s homo_sapiens -r 111
           ../bin/agfusion download -s mus_musculus -r 84
           pyensembl install --release 84 --species mus_musculus
           pyensembl install --release 75 --species homo_sapiens
-          pyensembl install --release 95 --species homo_sapiens
+          pyensembl install --release 111 --species homo_sapiens
       - name: Test with pytest
         run: |
           cd test

bin/build_dbs.py

@@ -0,0 +1,45 @@
+"""
+Download and build the agfusion database for a range of releases.
+"""
+
+import subprocess
+
+
+def run_command(command):
+    """Run a command and print it."""
+    print(f"Running: {command}")
+    subprocess.run(command, shell=True, check=True)
+
+
+species = "homo_sapiens"
+
+# Loop over releases from 96 to 110
+for i in range(96, 112):
+    # Check if the file exists on S3
+    s3_check_command = f"aws s3 ls s3://agfusion/agfusion.{species}.{i}.db.gz"
+    result = subprocess.run(
+        s3_check_command, shell=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE, check=True
+    )
+
+    if result.returncode == 0:
+        print(f"File for release {i} already exists in S3. Skipping...")
+        continue
+
+    # continue
+
+    print(f"Building release {i}")
+
+    # Install the release using pyensembl
+    run_command(f"pyensembl install --release {i} --species {species}")
+
+    # Build the agfusion database
+    run_command(f"agfusion build -d . -s {species} -r {i} --pfam Pfam-A.clans.tsv")
+
+    # Compress the database
+    run_command(f"gzip agfusion.{species}.{i}.db")
+
+    # Upload the compressed file to S3
+    run_command(f"aws s3 cp agfusion.{species}.{i}.db.gz s3://agfusion")
+
+    # Delete all files for the release from pyensembl
+    run_command(f"pyensembl delete-all-files --release {i} --species {species}")

test/test_parsers.py

@@ -1,5 +1,6 @@
 """Unit tests.
 """
+
 import unittest
 from os.path import abspath, curdir, join
 
@@ -15,9 +16,9 @@
 db_human = database.AGFusionDB(abspath(join(curdir, "agfusion.homo_sapiens.75.db")))
 db_human.build = "homo_sapiens_75"
 
-data_human95 = pyensembl.EnsemblRelease(95, "human")
-db_human95 = database.AGFusionDB(abspath(join(curdir, "agfusion.homo_sapiens.95.db")))
-db_human95.build = "homo_sapiens_95"
+data_human95 = pyensembl.EnsemblRelease(111, "human")
+db_human95 = database.AGFusionDB(abspath(join(curdir, "agfusion.homo_sapiens.111.db")))
+db_human95.build = "homo_sapiens_111"
 
 
 BASEDIR = "./data/FusionsFindingAlgorithms"