Update test_export.py

chanzuckerberg · Sep 10, 2024 · e9031fa · e9031fa
1 parent efb7012
commit e9031fa
Showing 1 changed file with 9 additions and 9 deletions.
diff --git a/src/backend/aspen/workflows/nextstrain_run/tests/test_export.py b/src/backend/aspen/workflows/nextstrain_run/tests/test_export.py
@@ -371,7 +371,7 @@ def test_targeted_config_simple(mocker, session, postgres_database, split_client
     subsampling_scheme = nextstrain_config["subsampling"][tree_type.value]
 
     # Just some placeholder sanity-checks
-    assert subsampling_scheme["closest"]["max_sequences"] == 100
+    assert subsampling_scheme["closest"]["max_sequences"] == 250
     assert subsampling_scheme["group"]["max_sequences"] == 50
     assert subsampling_scheme["state"]["max_sequences"] == 50
     assert subsampling_scheme["country"]["max_sequences"] == 25
@@ -430,7 +430,7 @@ def test_targeted_config_regions(mocker, session, postgres_database, split_clien
             )
 
         # Just some placeholder sanity-checks
-        assert subsampling_scheme["closest"]["max_sequences"] == 100
+        assert subsampling_scheme["closest"]["max_sequences"] == 250
         assert subsampling_scheme["group"]["max_sequences"] == 50
         assert subsampling_scheme["international"]["max_sequences"] == 100
         assert len(selected.splitlines()) == 10  # 5 gisaid samples + 5 selected samples
@@ -468,18 +468,18 @@ def test_targeted_config_large(mocker, session, postgres_database, split_client)
     mock_remote_db_uri(mocker, postgres_database.as_uri())
 
     tree_type = TreeType.TARGETED
-    phylo_run = create_test_data(session, split_client, tree_type, 200, 110, 10)
+    phylo_run = create_test_data(session, split_client, tree_type, 200, 270, 10)
     sequences, selected, metadata, nextstrain_config = generate_run(phylo_run.id)
 
     subsampling_scheme = nextstrain_config["subsampling"][tree_type.value]
 
     # Just some placeholder sanity-checks
-    assert subsampling_scheme["closest"]["max_sequences"] == 120
-    assert subsampling_scheme["group"]["max_sequences"] == 60
-    assert subsampling_scheme["state"]["max_sequences"] == 60
-    assert subsampling_scheme["country"]["max_sequences"] == 30
-    assert subsampling_scheme["international"]["max_sequences"] == 30
-    assert len(selected.splitlines()) == 120  # 10 gisaid samples + 110 selected samples
+    assert subsampling_scheme["closest"]["max_sequences"] == 280
+    assert subsampling_scheme["group"]["max_sequences"] == 140
+    assert subsampling_scheme["state"]["max_sequences"] == 140
+    assert subsampling_scheme["country"]["max_sequences"] == 70
+    assert subsampling_scheme["international"]["max_sequences"] == 70
+    assert len(selected.splitlines()) == 280  # 10 gisaid samples + 270 selected samples
     assert len(metadata.splitlines()) == 201  # 200 samples + 1 header line
     assert len(sequences.splitlines()) == 400  # 200 county samples, @2 lines each