Added save and restore of Optuna studies

notebooks/protac_degradation_predictor.py

@@ -783,24 +783,32 @@ def hyperparameter_tuning_and_training(
     # Create an Optuna study object
     sampler = TPESampler(seed=42, multivariate=True)
     study = optuna.create_study(direction='minimize', sampler=sampler)
-    study.optimize(
-        lambda trial: objective(
-            trial,
-            train_df,
-            val_df,
-            hidden_dim_options=hidden_dim_options,
-            batch_size_options=batch_size_options,
-            learning_rate_options=learning_rate_options,
-            smote_k_neighbors_options=smote_k_neighbors_options,
-            fast_dev_run=fast_dev_run,
-            active_label=active_label,
-            disabled_embeddings=disabled_embeddings,
-        ),
-        n_trials=n_trials,
-    )
 
+    study_loaded = False
     if study_filename:
-        joblib.dump(study, study_filename)
+        if os.path.exists(study_filename):
+            study = joblib.load(study_filename)
+            study_loaded = True
+            print(f'Loaded study from {study_filename}')
+
+    if not study_loaded:
+        study.optimize(
+            lambda trial: objective(
+                trial,
+                train_df,
+                val_df,
+                hidden_dim_options=hidden_dim_options,
+                batch_size_options=batch_size_options,
+                learning_rate_options=learning_rate_options,
+                smote_k_neighbors_options=smote_k_neighbors_options,
+                fast_dev_run=fast_dev_run,
+                active_label=active_label,
+                disabled_embeddings=disabled_embeddings,
+            ),
+            n_trials=n_trials,
+        )
+        if study_filename:
+            joblib.dump(study, study_filename)
 
     # Retrain the model with the best hyperparameters
     model, trainer, metrics = train_model(
@@ -1030,7 +1038,7 @@ def main(
             del trainer
 
             # Ablation study: disable embeddings at a time
-            for disabled_embeddings in [['poi'], ['cell'], ['smiles'], ['e3', 'cell'], ['poi', 'e3', 'cell']]:
+            for disabled_embeddings in [['e3'], ['poi'], ['cell'], ['smiles'], ['e3', 'cell'], ['poi', 'e3', 'cell']]:
                 print('-' * 100)
                 print(f'Ablation study with disabled embeddings: {disabled_embeddings}')
                 print('-' * 100)

reports/study_Active_Dmax_0.6_pDC50_6.0_e3_ligase_fold_0_test_split_0.1.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:258ead80bfa66c98f250374143a4587f4b8e51d6449bbeb54100ae06e3ab1b91
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_e3_ligase_fold_0_test_split_0.2.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5329684c4b6e305284aadbec651c0ad44fcee795f0a09c15b1c395150596ea1f
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_e3_ligase_fold_1_test_split_0.1.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5e6a65a3dfed7f5281a55f97619e94b44e7a4ecc0ed20d97c150ce583dcdd220
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_e3_ligase_fold_1_test_split_0.2.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6366b23ad7b3f37c135efe546e9c2fa81a697a6abac605116d7b3178265c8c46
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_e3_ligase_fold_2_test_split_0.1.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0d245147130f8303d37e9995b35db618c480f8679b97e12c4d069f90b3b806c
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_e3_ligase_fold_2_test_split_0.2.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:41abf93de0f3d7c3766c4df0b5ad9a95b0f84ad7acb145ab416d9058dbcb3972
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_e3_ligase_fold_3_test_split_0.1.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cf374269d482f85afadd83b0bbf3bb8cc4d600d07711779c9edf94676f0d32fc
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_e3_ligase_fold_3_test_split_0.2.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6a0386512fb6a9aae07562a8ab6f1a87471969153105461ebd509420a1f59b66
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_e3_ligase_fold_4_test_split_0.1.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5d9b5b42bdb3140bbfc45e94022e92464b87f328789e490e64ea9dacb6905100
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_e3_ligase_fold_4_test_split_0.2.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:42bda2152c32dd94070c523ff5dfb976a422038cda44bf32fcb941c897aec467
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_random_fold_0_test_split_0.1.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0fc7210ba0e68361426dea9f6663e20c23997e88146436dc8a26948c18b97e1f
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_random_fold_0_test_split_0.2.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e490b0ed7440b8ea51ff265d8a68a1b0f3d1cea260461eb47dbeac92430c42bb
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_random_fold_1_test_split_0.1.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cb5fb7244b181e7fe1754570bc8878d413e3c03957e497be3e82042bca49c181
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_random_fold_1_test_split_0.2.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:53792a857d41cf5f2f19c17a000fddc9709990c831ae5d99380a4dc1c44f70be
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_random_fold_2_test_split_0.1.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ea9e6e8e475ef6d0a37b534086aea24ce2e7d7ca7fd7146472d09066d19f3888
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_random_fold_2_test_split_0.2.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f137f1f60be4f7ab3f88c945fc4cb42e0c65750d70c4cf75cfe93881f2f0c1d9
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_random_fold_3_test_split_0.1.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ab8b97e70915fb864e26c086833306d40fe4e64bd4d46fc95fc67551eac31d51
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_random_fold_3_test_split_0.2.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a51f48587376a0e60b62633193c301a45e14e27409f9453f1c33088863a158d7
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_random_fold_4_test_split_0.1.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:95cf6ff3653506b6118c14a4738c1bc5c87969309ae7052fb794fdecf1a59550
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_random_fold_4_test_split_0.2.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:73a4684a20c4ab2a7cbe7fac42b6c48d154a7643031da38d35ca0e2dbce98e82
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_tanimoto_fold_0_test_split_0.1.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:27aa74ab92272f4c455f8eb32e7da3d3b71e213937f9a96a7d07e3ca61af06fb
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_tanimoto_fold_0_test_split_0.2.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:647f8b8e72f9f1f72ecdb8733b306df290f62fee42d30ab6da0f26cf3ed3b010
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_tanimoto_fold_1_test_split_0.1.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e8b8cd9536d7d1fab755506dc6cddf5d0d66ae3743f3a687dc5e82d44b134cd7
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_tanimoto_fold_1_test_split_0.2.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:375e5f4f5080d2cf654b932faedeb2b0e9433d0c738542098f789beecd980c65
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_tanimoto_fold_2_test_split_0.1.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2503fafc20807b89d6663a978a1a52d923cd36a43e09b2cd5761bfd42c505942
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_tanimoto_fold_2_test_split_0.2.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f7298a04d0888f4de87041efd6b78c42e13c3f1630c43567d582bc7710a40847
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_tanimoto_fold_3_test_split_0.1.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ac2296caf215b91d3afbf2f31a625cdade8a1557f791e43350fe86e04373c6f7
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_tanimoto_fold_3_test_split_0.2.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d2a7a9f6ed11e1b5b6f876dd927612d03f4780f9db3e65b9f1ebb8fbd853677f
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_tanimoto_fold_4_test_split_0.1.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:994f1b499f359d0fb32eb708054826edecd317862cff181455fae8040330f6b9
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_tanimoto_fold_4_test_split_0.2.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fd4c40033da16a1cee16fd998c82e0403a31db4d14ba0604160ea143bae03668
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_uniprot_fold_0_test_split_0.1.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:24bd2673efb0d3855509d2cd33f08f2d455994a63663300225753bc87f8067c1
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_uniprot_fold_0_test_split_0.2.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6d6f639ff3a4d6c1b73388bf02f2bfaea087a5dd8eb5ce60080276039cbf08ad
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_uniprot_fold_1_test_split_0.1.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:613e1eb1538067c795baf80a6ec8fc3fd81f952b34cf119a9b21ba0e0fa0db73
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_uniprot_fold_1_test_split_0.2.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:85569ae5c9c134baa736d5235e7b62466e92abc364c8ba95d8c995f06a589486
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_uniprot_fold_2_test_split_0.1.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:20ab1d27b21e5afda52c2a13ba152e1b5a3b9ad471b5e453ed48809be6721271
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_uniprot_fold_2_test_split_0.2.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:63a3b9252d05935811200a76ae7de1bcc3e90ff058082f77cd5dfc082e697187
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_uniprot_fold_3_test_split_0.1.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ba80e55724bab9860cd417bd6a6b9047a40a572a7e2ffc8e3d52af8e61036a7f
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_uniprot_fold_3_test_split_0.2.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e71cf47ccba99e2571a0d761016c7374cb894914a9fdb0aa68b8bdac98801d93
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_uniprot_fold_4_test_split_0.1.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7e2f1a70f7a8ec861d039db65f3ba91458d08fdda3887c2a6ecb9e91bbb4f514
+size 45164

reports/study_Active_Dmax_0.6_pDC50_6.0_uniprot_fold_4_test_split_0.2.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:79ead33aaa6246c3125a27ca452703b798d0f89df0887e9cad1e1c8907ff2fbd
+size 45164