Fix no space left on device error in CI during docker build (#82)

* fix: avoid installing nvidia libs

* fix: alternative approach for torch installation

* docs: ping src to trigger ci

* feat: try fail test handling CI

* update & cleanup test_data

* bugfix: restore system loading test - update test system_list

* chore: update test entry .json !!!

---------

Co-authored-by: OleinikovasV <[email protected]>

.github/workflows/pr.yaml

@@ -65,13 +65,20 @@ jobs:
         if: steps.changed-src.outputs.any_changed == 'true' && steps.changed-base.outputs.any_changed != 'true'
         run: python flows/docker.py pull
       - name: Build image
-        if: steps.changed-src.outputs.any_changed == 'true'
+        if: steps.changed-base.outputs.any_changed == 'true' || steps.changed-src.outputs.any_changed == 'true'
         run: python flows/docker.py build
       - name: Run test image
-        if: steps.changed-src.outputs.any_changed == 'true'
-        run: python flows/docker.py test
+        if: steps.changed-base.outputs.any_changed == 'true' || steps.changed-src.outputs.any_changed == 'true'
+        run: |
+          python flows/docker.py test
+          if ! test -f reports/.coverage; then
+            echo "Missing test coverage reports (reports/.coverage). Not all tests have passed?"
+            exit 1
+          else
+            echo "Test coverage report found!"
+          fi
       - name: Copy and surgery coverage
-        if: steps.changed-src.outputs.any_changed == 'true'
+        if: steps.changed-base.outputs.any_changed == 'true' || steps.changed-src.outputs.any_changed == 'true'
         run: |
           cp reports/.coverage .
           sqlite3 .coverage "update file set path='src/' || substr(path, 40);"

dockerfiles/base/requirements.txt

@@ -19,7 +19,7 @@ async-timeout==4.0.3      # via aiohttp
 attrs==24.2.0             # via aiohttp, jsonschema, pytest, referencing
 babel==2.16.0             # via jupyterlab-server
 beautifulsoup4==4.12.3    # via nbconvert
-biotite==0.39.0           # via plinder (pyproject.toml)
+biotite==1.0.1            # via plinder (pyproject.toml)
 bleach==6.1.0             # via nbconvert
 build==0.9.0              # via plinder (pyproject.toml)
 cachetools==5.5.0         # via google-auth, tox

requirements_data.txt

@@ -2,4 +2,4 @@
   tabulate
   pdb-validation @ git+https://git.scicore.unibas.ch/schwede/ligand-validation.git
   mmpdb @ git+https://github.com/rdkit/mmpdb.git
-  torch
+  https://download.pytorch.org/whl/cpu/torch-2.5.1%2Bcpu-cp310-cp310-linux_x86_64.whl#sha256=7f91a2200e352745d70e22396bd501448e28350fbdbd8d8b1c83037e25451150

src/plinder/core/utils/gcs.py

@@ -72,7 +72,7 @@ def download_as_str(
 ) -> str:
     assert bucket is not None
     blob = bucket.blob("/".join(Path(gcs_path).parts[2:]))
-    return str(blob.download_as_string().decode("utf8"))
+    return str(blob.download_as_bytes().decode("utf8"))
 
 
 @retry

src/plinder/data/pipeline/pipeline.py

@@ -17,8 +17,7 @@ class IngestPipeline:
         - scatter
         - compute
         - join
-    by convention in method names. This could be
-    enforced with a metaclass construct but not enough time.
+    by convention in method names.
 
     scatter methods return lists of lists of primitives
     compute methods may return something if intended to be joined

tests/core/test_core_system.py

@@ -10,14 +10,17 @@
     [
         "19hc__1__1.A_1.B__1.D_1.L_1.Q_1.S_1.U",
         "19hc__1__1.A_1.B__1.E_1.F_1.H_1.J_1.O",
+        "19hc__1__1.A_1.B__1.G",
         "19hc__1__1.A_1.B__1.K_1.M_1.N",
+        "19hc__1__1.A_1.B__1.R",
         "19hc__1__1.A_1.B__1.V_1.X_1.Y",
-        "19hc__1__1.A__1.G",
+        "19hc__1__1.A_1.B__1.W",
         "19hc__1__1.A__1.I",
+        "19hc__1__1.B__1.T",
     ],
 )
 def test_plinder_system(system_id, read_plinder_mount):
-    index.PlinderSystem(system_id=system_id)
+    index.PlinderSystem(system_id=system_id).system
 
 
 @pytest.mark.parametrize(
@@ -35,7 +38,7 @@ def test_plinder_system_fails(system_id, read_plinder_mount):
 def test_plinder_system_system_files(read_plinder_mount):
     system_id = "19hc__1__1.A_1.B__1.V_1.X_1.Y"
     s = index.PlinderSystem(system_id=system_id)
-    assert len(s.structures) == 10
+    assert len(s.structures) == 9
     assert len(s.ligand_sdfs) == 3
     assert len(s.system_cif)
     assert len(s.receptor_cif)

tests/core/test_gcs.py

@@ -11,7 +11,7 @@ class _Blob:
     def __init__(self, name):
         self.name = name
 
-    def download_as_string(self):
+    def download_as_bytes(self):
         return b"test"
 
     def download_to_filename(self, local_path):

tests/core/test_superimpose.py

@@ -8,7 +8,9 @@ def test_superimpose_chain(read_plinder_mount):
     In all cases the superimposed structure should have the original number of atoms
     and a low RMSD to the fixed structure.
     """
-    system_id_1 = "19hc__1__1.A__1.G"
+    # TODO: review if this test is still relevant
+    pass
+    system_id_1 = "19hc__1__1.A_1.B__1.G"
     system_id_2 = "19hc__1__1.A_1.B__1.V_1.X_1.Y"
     # system_dir_1 = read_plinder_mount / "systems" / system_id_1
     # system_dir_2 = read_plinder_mount / "systems" / system_id_2
@@ -20,12 +22,18 @@ def test_superimpose_chain(read_plinder_mount):
     chain_1_array = struct1.protein_atom_array[
         struct1.protein_atom_array.chain_id == chain_id_1
     ]
-    # TODO: test assertions here
-    # chain_2_array = sys_2.protein_atom_array[
-    #     sys_2.protein_atom_array.chain_id == chain_id_2
+    # # TODO: test assertions here
+    # chain_2_array = struct2.protein_atom_array[
+    #     struct2.protein_atom_array.chain_id == chain_id_2
     # ]
+
     super_chain_1, raw_rmsd, refined_rmsd = struct1.superimpose(struct2)
     assert isinstance(super_chain_1, Structure)
-    assert super_chain_1.protein_atom_array.shape == chain_1_array.shape
-    assert (raw_rmsd == 0 and refined_rmsd == 0) or raw_rmsd > refined_rmsd
+    super_chain_1_array = super_chain_1.protein_atom_array[
+        super_chain_1.protein_atom_array.chain_id == chain_id_1
+    ]
+    assert super_chain_1_array.shape == chain_1_array.shape
+
+    # check alignment quality
+    assert abs(raw_rmsd - refined_rmsd) < 0.01 or raw_rmsd > refined_rmsd
     assert refined_rmsd < 2.0