update: update file paths to include 'notebooks' directory for interaction dataframes

Author: amorehead

notebooks/astex_method_interaction_analysis_plotting_slurm.py

@@ -200,7 +200,7 @@ def signal_handler(signum, frame):
 # ##### Analyze `Astex Diverse` set interactions as a baseline
 
 # %%
-if not os.path.exists("astex_diverse_interaction_dataframes.h5"):
+if not os.path.exists(os.path.join("notebooks", "astex_diverse_interaction_dataframes.h5")):
     ad_protein_ligand_filepath_pairs = []
     for item in os.listdir(ad_set_dir):
         ligand_item_path = os.path.join(ad_set_dir, item)
@@ -235,7 +235,7 @@ def signal_handler(signum, frame):
             continue
 
         # NOTE: we iteratively save the interaction dataframes to an HDF5 file
-        with pd.HDFStore("astex_diverse_interaction_dataframes.h5") as store:
+        with pd.HDFStore(os.path.join("notebooks", "astex_diverse_interaction_dataframes.h5")) as store:
             for i, df in enumerate(ad_protein_ligand_interaction_dfs):
                 store.put(f"df_{i}", df)
 
@@ -282,7 +282,7 @@ def signal_handler(signum, frame):
         method = method.split("_")[0]
 
         if not os.path.exists(
-            f"{method}{single_seq_suffix}{vina_suffix}_{dataset}_interaction_dataframes_{repeat_index}.h5"
+            os.path.join("notebooks", f"{method}{single_seq_suffix}{vina_suffix}_{dataset}_interaction_dataframes_{repeat_index}.h5")
         ):
             with open_dict(cfg):
                 cfg.method = method
@@ -358,7 +358,7 @@ def signal_handler(signum, frame):
 
                 # NOTE: we iteratively save the interaction dataframes to an HDF5 file
                 with pd.HDFStore(
-                    f"{method}{single_seq_suffix}{vina_suffix}_{dataset}_interaction_dataframes_{repeat_index}.h5"
+                    os.path.join("notebooks", f"{method}{single_seq_suffix}{vina_suffix}_{dataset}_interaction_dataframes_{repeat_index}.h5")
                 ) as store:
                     for i, df in enumerate(astex_protein_ligand_interaction_dfs):
                         store.put(f"df_{i}", df)
@@ -413,12 +413,12 @@ def process_method(file_path, category):
 for method in baseline_methods:
     for repeat_index in range(1, max_num_repeats_per_method + 1):
         method_title = method_mapping[method]
-        file_path = f"{method}_astex_diverse_interaction_dataframes_{repeat_index}.h5"
+        file_path = os.path.join("notebooks", f"{method}_astex_diverse_interaction_dataframes_{repeat_index}.h5")
         if os.path.exists(file_path):
             dfs.append(process_method(file_path, method_title))
 
-if os.path.exists("astex_diverse_interaction_dataframes.h5"):
-    dfs.append(process_method("astex_diverse_interaction_dataframes.h5", "Reference"))
+if os.path.exists(os.path.join("notebooks", "astex_diverse_interaction_dataframes.h5")):
+    dfs.append(process_method(os.path.join("notebooks", "astex_diverse_interaction_dataframes.h5"), "Reference"))
 
 # combine statistics
 assert len(dfs) > 0, "No interaction dataframes found."
@@ -539,14 +539,14 @@ def histogram_to_vector(histogram, bins):
 for method in baseline_methods:
     for repeat_index in range(1, max_num_repeats_per_method + 1):
         method_title = method_mapping[method]
-        file_path = f"{method}_astex_diverse_interaction_dataframes_{repeat_index}.h5"
+        file_path = os.path.join("notebooks", f"{method}_astex_diverse_interaction_dataframes_{repeat_index}.h5")
         if os.path.exists(file_path):
             dfs.append(bin_interactions(file_path, method_title))
 
 assert os.path.exists(
-    "astex_diverse_interaction_dataframes.h5"
+    os.path.join("notebooks", "astex_diverse_interaction_dataframes.h5")
 ), "No reference interaction dataframe found."
-reference_df = bin_interactions("astex_diverse_interaction_dataframes.h5", "Reference")
+reference_df = bin_interactions(os.path.join("notebooks", "astex_diverse_interaction_dataframes.h5"), "Reference")
 
 # combine bins from all method dataframes
 assert len(dfs) > 0, "No interaction dataframes found."

notebooks/casp15_method_interaction_analysis_plotting_slurm.py

@@ -200,7 +200,7 @@ def signal_handler(signum, frame):
 # ##### Analyze `CASP15` set interactions as a baseline
 
 # %%
-if not os.path.exists("casp15_interaction_dataframes.h5"):
+if not os.path.exists(os.path.join("notebooks", "casp15_interaction_dataframes.h5")):
     casp15_protein_ligand_complex_filepaths = []
     for item in os.listdir(casp15_set_dir):
         item_path = os.path.join(casp15_set_dir, item)
@@ -244,7 +244,7 @@ def signal_handler(signum, frame):
             continue
 
         # NOTE: we iteratively save the interaction dataframes to an HDF5 file
-        with pd.HDFStore("casp15_interaction_dataframes.h5") as store:
+        with pd.HDFStore(os.path.join("notebooks", "casp15_interaction_dataframes.h5")) as store:
             for i, df in enumerate(casp15_protein_ligand_interaction_dfs):
                 store.put(f"df_{i}", df)
 
@@ -261,7 +261,7 @@ def signal_handler(signum, frame):
     for repeat_index in range(1, max_num_repeats_per_method + 1):
         method_title = method_mapping[method]
 
-        if not os.path.exists(f"{method}_{dataset}_interaction_dataframes_{repeat_index}.h5"):
+        if not os.path.exists(os.path.join("notebooks", f"{method}_{dataset}_interaction_dataframes_{repeat_index}.h5")):
             method_casp15_set_dir = os.path.join(
                 "data",
                 "test_cases",
@@ -340,7 +340,7 @@ def signal_handler(signum, frame):
 
                 # NOTE: we iteratively save the interaction dataframes to an HDF5 file
                 with pd.HDFStore(
-                    f"{method}_{dataset}_interaction_dataframes_{repeat_index}.h5"
+                    os.path.join("notebooks", f"{method}_{dataset}_interaction_dataframes_{repeat_index}.h5")
                 ) as store:
                     for i, df in enumerate(casp15_protein_ligand_interaction_dfs):
                         store.put(f"df_{i}", df)
@@ -395,12 +395,12 @@ def process_method(file_path, category):
 for method in baseline_methods:
     for repeat_index in range(1, max_num_repeats_per_method + 1):
         method_title = method_mapping[method]
-        file_path = f"{method}_casp15_interaction_dataframes_{repeat_index}.h5"
+        file_path = os.path.join("notebooks", f"{method}_casp15_interaction_dataframes_{repeat_index}.h5")
         if os.path.exists(file_path):
             dfs.append(process_method(file_path, method_title))
 
-if os.path.exists("casp15_interaction_dataframes.h5"):
-    dfs.append(process_method("casp15_interaction_dataframes.h5", "Reference"))
+if os.path.exists(os.path.join("notebooks", "casp15_interaction_dataframes.h5")):
+    dfs.append(process_method(os.path.join("notebooks", "casp15_interaction_dataframes.h5"), "Reference"))
 
 # combine statistics
 assert len(dfs) > 0, "No interaction dataframes found."
@@ -521,14 +521,14 @@ def histogram_to_vector(histogram, bins):
 for method in baseline_methods:
     for repeat_index in range(1, max_num_repeats_per_method + 1):
         method_title = method_mapping[method]
-        file_path = f"{method}_casp15_interaction_dataframes_{repeat_index}.h5"
+        file_path = os.path.join("notebooks", f"{method}_casp15_interaction_dataframes_{repeat_index}.h5")
         if os.path.exists(file_path):
             dfs.append(bin_interactions(file_path, method_title))
 
 assert os.path.exists(
-    "casp15_interaction_dataframes.h5"
+    os.path.join("notebooks", "casp15_interaction_dataframes.h5")
 ), "No reference interaction dataframe found."
-reference_df = bin_interactions("casp15_interaction_dataframes.h5", "Reference")
+reference_df = bin_interactions(os.path.join("notebooks", "casp15_interaction_dataframes.h5"), "Reference")
 
 # combine bins from all method dataframes
 assert len(dfs) > 0, "No interaction dataframes found."

notebooks/dockgen_method_interaction_analysis_plotting_slurm.py

@@ -200,7 +200,7 @@ def signal_handler(signum, frame):
 # ##### Analyze `DockGen` set interactions as a baseline
 
 # %%
-if not os.path.exists("dockgen_interaction_dataframes.h5"):
+if not os.path.exists(os.path.join("notebooks", "dockgen_interaction_dataframes.h5")):
     dockgen_test_ids_filepath = os.path.join(
         "data", "dockgen_set", "split_test.txt"
     )  # NOTE: change as needed
@@ -248,7 +248,7 @@ def signal_handler(signum, frame):
             continue
 
         # NOTE: we iteratively save the interaction dataframes to an HDF5 file
-        with pd.HDFStore("dockgen_interaction_dataframes.h5") as store:
+        with pd.HDFStore(os.path.join("notebooks", "dockgen_interaction_dataframes.h5")) as store:
             for i, df in enumerate(dg_protein_ligand_interaction_dfs):
                 store.put(f"df_{i}", df)
 
@@ -294,9 +294,7 @@ def signal_handler(signum, frame):
 
         method = method.split("_")[0]
 
-        if not os.path.exists(
-            f"{method}{single_seq_suffix}{vina_suffix}_{dataset}_interaction_dataframes_{repeat_index}.h5"
-        ):
+        if not os.path.exists(os.path.join("notebooks", f"{method}{single_seq_suffix}{vina_suffix}_{dataset}_interaction_dataframes_{repeat_index}.h5")):
             with open_dict(cfg):
                 cfg.method = method
                 cfg.repeat_index = repeat_index
@@ -371,7 +369,7 @@ def signal_handler(signum, frame):
 
                 # NOTE: we iteratively save the interaction dataframes to an HDF5 file
                 with pd.HDFStore(
-                    f"{method}{single_seq_suffix}{vina_suffix}_{dataset}_interaction_dataframes_{repeat_index}.h5"
+                    os.path.join("notebooks", f"{method}{single_seq_suffix}{vina_suffix}_{dataset}_interaction_dataframes_{repeat_index}.h5")
                 ) as store:
                     for i, df in enumerate(dockgen_protein_ligand_interaction_dfs):
                         store.put(f"df_{i}", df)
@@ -426,12 +424,12 @@ def process_method(file_path, category):
 for method in baseline_methods:
     for repeat_index in range(1, max_num_repeats_per_method + 1):
         method_title = method_mapping[method]
-        file_path = f"{method}_dockgen_interaction_dataframes_{repeat_index}.h5"
+        file_path = os.path.join("notebooks", f"{method}_dockgen_interaction_dataframes_{repeat_index}.h5")
         if os.path.exists(file_path):
             dfs.append(process_method(file_path, method_title))
 
-if os.path.exists("dockgen_interaction_dataframes.h5"):
-    dfs.append(process_method("dockgen_interaction_dataframes.h5", "Reference"))
+if os.path.exists(os.path.join("notebooks", "dockgen_interaction_dataframes.h5")):
+    dfs.append(process_method(os.path.join("notebooks", "dockgen_interaction_dataframes.h5"), "Reference"))
 
 # combine statistics
 assert len(dfs) > 0, "No interaction dataframes found."
@@ -552,14 +550,14 @@ def histogram_to_vector(histogram, bins):
 for method in baseline_methods:
     for repeat_index in range(1, max_num_repeats_per_method + 1):
         method_title = method_mapping[method]
-        file_path = f"{method}_dockgen_interaction_dataframes_{repeat_index}.h5"
+        file_path = os.path.join("notebooks", f"{method}_dockgen_interaction_dataframes_{repeat_index}.h5")
         if os.path.exists(file_path):
             dfs.append(bin_interactions(file_path, method_title))
 
 assert os.path.exists(
-    "dockgen_interaction_dataframes.h5"
+    os.path.join("notebooks", "dockgen_interaction_dataframes.h5")
 ), "No reference interaction dataframe found."
-reference_df = bin_interactions("dockgen_interaction_dataframes.h5", "Reference")
+reference_df = bin_interactions(os.path.join("notebooks", "dockgen_interaction_dataframes.h5"), "Reference")
 
 # combine bins from all method dataframes
 assert len(dfs) > 0, "No interaction dataframes found."

notebooks/posebusters_method_interaction_analysis_plotting_slurm.py

@@ -219,7 +219,7 @@ def signal_handler(signum, frame):
 # ##### Analyze `PoseBusters Benchmark` set interactions as a baseline
 
 # %%
-if not os.path.exists("posebusters_benchmark_interaction_dataframes.h5"):
+if not os.path.exists(os.path.join("notebooks", "posebusters_benchmark_interaction_dataframes.h5")):
     posebusters_ccd_ids_filepath = os.path.join(
         "data",
         "posebusters_pdb_ccd_ids.txt",
@@ -265,7 +265,7 @@ def signal_handler(signum, frame):
             continue
 
         # NOTE: we iteratively save the interaction dataframes to an HDF5 file
-        with pd.HDFStore("posebusters_benchmark_interaction_dataframes.h5") as store:
+        with pd.HDFStore(os.path.join("notebooks", "posebusters_benchmark_interaction_dataframes.h5")) as store:
             for i, df in enumerate(pb_protein_ligand_interaction_dfs):
                 store.put(f"df_{i}", df)
 
@@ -312,7 +312,7 @@ def signal_handler(signum, frame):
         method = method.split("_")[0]
 
         if not os.path.exists(
-            f"{method}{single_seq_suffix}{vina_suffix}_{dataset}_interaction_dataframes_{repeat_index}.h5"
+            os.path.join("notebooks", f"{method}{single_seq_suffix}{vina_suffix}_{dataset}_interaction_dataframes_{repeat_index}.h5")
         ):
             with open_dict(cfg):
                 cfg.method = method
@@ -390,7 +390,7 @@ def signal_handler(signum, frame):
 
                 # NOTE: we iteratively save the interaction dataframes to an HDF5 file
                 with pd.HDFStore(
-                    f"{method}{single_seq_suffix}{vina_suffix}_{dataset}_interaction_dataframes_{repeat_index}.h5"
+                    os.path.join("notebooks", f"{method}{single_seq_suffix}{vina_suffix}_{dataset}_interaction_dataframes_{repeat_index}.h5")
                 ) as store:
                     for i, df in enumerate(posebusters_protein_ligand_interaction_dfs):
                         store.put(f"df_{i}", df)
@@ -452,12 +452,12 @@ def process_method(file_path, category):
 for method in baseline_methods:
     for repeat_index in range(1, max_num_repeats_per_method + 1):
         method_title = method_mapping[method]
-        file_path = f"{method}_posebusters_benchmark_interaction_dataframes_{repeat_index}.h5"
+        file_path = os.path.join("notebooks", f"{method}_posebusters_benchmark_interaction_dataframes_{repeat_index}.h5")
         if os.path.exists(file_path):
             dfs.append(process_method(file_path, method_title))
 
-if os.path.exists("posebusters_benchmark_interaction_dataframes.h5"):
-    dfs.append(process_method("posebusters_benchmark_interaction_dataframes.h5", "Reference"))
+if os.path.exists(os.path.join("notebooks", "posebusters_benchmark_interaction_dataframes.h5")):
+    dfs.append(process_method(os.path.join("notebooks", "posebusters_benchmark_interaction_dataframes.h5"), "Reference"))
 
 # combine statistics
 assert len(dfs) > 0, "No interaction dataframes found."
@@ -581,14 +581,15 @@ def histogram_to_vector(histogram, bins):
 for method in baseline_methods:
     for repeat_index in range(1, max_num_repeats_per_method + 1):
         method_title = method_mapping[method]
-        file_path = f"{method}_posebusters_benchmark_interaction_dataframes_{repeat_index}.h5"
+        file_path = os.path.join("notebooks", f"{method}_posebusters_benchmark_interaction_dataframes_{repeat_index}.h5")
         if os.path.exists(file_path):
             dfs.append(bin_interactions(file_path, method_title))
 
 assert os.path.exists(
-    "posebusters_benchmark_interaction_dataframes.h5"
+    os.path.join("notebooks", "posebusters_benchmark_interaction_dataframes.h5")
+
 ), "No reference interaction dataframe found."
-reference_df = bin_interactions("posebusters_benchmark_interaction_dataframes.h5", "Reference")
+reference_df = bin_interactions(os.path.join("notebooks", "posebusters_benchmark_interaction_dataframes.h5"), "Reference")
 
 # combine bins from all method dataframes
 assert len(dfs) > 0, "No interaction dataframes found."