Update notebook

BioinfoMachineLearning · Feb 3, 2025 · 80aa109 · 80aa109
1 parent 47bbb92
commit 80aa109
Showing 1 changed file with 29 additions and 4 deletions.
diff --git a/notebooks/failure_modes_analysis_plotting.ipynb b/notebooks/failure_modes_analysis_plotting.ipynb
@@ -1378,6 +1378,7 @@
     "\n",
     "failed_complexes_to_study_further = []\n",
     "failed_complex_pdb_ids_to_study_further = []\n",
+    "failed_complex_function_annotations_to_study_further = []\n",
     "for complex_pdb_id in collated_complexes:\n",
     "    collated_complex_chains = collated_complexes[complex_pdb_id]\n",
     "    all_chains_after_cutoff_deposition_date = True\n",
@@ -1393,22 +1394,46 @@
     "            all_chains_after_cutoff_deposition_date = False\n",
     "            break\n",
     "    if all_chains_after_cutoff_deposition_date:\n",
+    "        complex_pdb_id_ = complex_pdb_id.lower().split(\"_\")[0]\n",
+    "        if complex_pdb_id_ == \"?\":\n",
+    "            continue\n",
+    "        if complex_pdb_id_ in pdb_info_cache:\n",
+    "            complex_pdb_info = pdb_info_cache[complex_pdb_id_]\n",
+    "        else:\n",
+    "            complex_pdb_info = pypdb.get_all_info(complex_pdb_id_)\n",
+    "            pdb_info_cache[complex_pdb_id_] = complex_pdb_info\n",
+    "        if not complex_pdb_info:\n",
+    "            continue\n",
+    "        complex_function_annotation = (\n",
+    "            # NOTE: these represent functional keywords\n",
+    "            complex_pdb_info[\"struct_keywords\"][\"pdbx_keywords\"]\n",
+    "            .lower()\n",
+    "            .split(\", \")[0]\n",
+    "        )\n",
     "        failed_complexes_to_study_further.append(collated_complex_chains)\n",
     "        failed_complex_pdb_ids_to_study_further.append(complex_pdb_id)\n",
-    "        print(f\"{complex_pdb_id}: {collated_complex_chains}\")\n",
+    "        failed_complex_function_annotations_to_study_further.append(complex_function_annotation)\n",
+    "        print(f\"{complex_pdb_id}, {complex_function_annotation} : {collated_complex_chains}\")\n",
     "\n",
-    "for complex_index, complex_pdb_id in enumerate(failed_complex_pdb_ids_to_study_further):\n",
+    "for complex_index, (complex_pdb_id, complex_function_annotation) in enumerate(\n",
+    "    zip(\n",
+    "        failed_complex_pdb_ids_to_study_further,\n",
+    "        failed_complex_function_annotations_to_study_further,\n",
+    "    )\n",
+    "):\n",
     "    if complex_index == 0:\n",
     "        print()\n",
     "    pb_path = os.path.join(\n",
     "        \"..\", \"data\", \"posebusters_benchmark_set\", f\"{complex_pdb_id.upper()}_*\", \"*.pdb\"\n",
     "    )\n",
     "    if glob.glob(pb_path):\n",
     "        pb_path = glob.glob(pb_path)[0]\n",
-    "        print(f\"Posebusters Benchmark target: {complex_pdb_id} -> {pb_path}\")\n",
+    "        print(\n",
+    "            f\"Posebusters Benchmark target: {complex_pdb_id}, {complex_function_annotation} -> {pb_path}\"\n",
+    "        )\n",
     "    else:\n",
     "        casp_matches = [k for k, v in casp15_target_pdb_id_mapping.items() if v == complex_pdb_id]\n",
-    "        print(f\"CASP15 target: {complex_pdb_id} -> {casp_matches}\")\n",
+    "        print(f\"CASP15 target: {complex_pdb_id}, {complex_function_annotation} -> {casp_matches}\")\n",
     "\n",
     "print(\n",
     "    f\"{len(failed_complexes_to_study_further)} ((failed complex, deposition date), (top match, deposition date)) novel (multi-chain) protein-ligand PDB complexes AlphaFold 3 failed to predict that are worth studying further.\"\n",