Push RetroTAC ensemble

.gitattributes

@@ -33,3 +33,24 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+members/mlp_seed1011_fold2.skops filter=lfs diff=lfs merge=lfs -text
+members/mlp_seed1011_fold3.skops filter=lfs diff=lfs merge=lfs -text
+members/mlp_seed1011_fold4.skops filter=lfs diff=lfs merge=lfs -text
+members/mlp_seed42_fold2.skops filter=lfs diff=lfs merge=lfs -text
+members/mlp_seed456_fold0.skops filter=lfs diff=lfs merge=lfs -text
+members/mlp_seed456_fold1.skops filter=lfs diff=lfs merge=lfs -text
+members/mlp_seed456_fold2.skops filter=lfs diff=lfs merge=lfs -text
+members/xgb_seed1011_fold3.skops filter=lfs diff=lfs merge=lfs -text
+members/xgb_seed1011_fold3.ubj filter=lfs diff=lfs merge=lfs -text
+members/xgb_seed1011_fold4.skops filter=lfs diff=lfs merge=lfs -text
+members/xgb_seed1011_fold4.ubj filter=lfs diff=lfs merge=lfs -text
+members/xgb_seed42_fold2.skops filter=lfs diff=lfs merge=lfs -text
+members/xgb_seed42_fold2.ubj filter=lfs diff=lfs merge=lfs -text
+members/xgb_seed42_fold4.skops filter=lfs diff=lfs merge=lfs -text
+members/xgb_seed42_fold4.ubj filter=lfs diff=lfs merge=lfs -text
+members/xgb_seed456_fold4.skops filter=lfs diff=lfs merge=lfs -text
+members/xgb_seed456_fold4.ubj filter=lfs diff=lfs merge=lfs -text
+members/xgb_seed789_fold1.skops filter=lfs diff=lfs merge=lfs -text
+members/xgb_seed789_fold1.ubj filter=lfs diff=lfs merge=lfs -text
+members/xgb_seed789_fold2.skops filter=lfs diff=lfs merge=lfs -text
+members/xgb_seed789_fold2.ubj filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,19 @@
+# ailab-bio/RetroTAC
+
+Caruana-weighted ensemble of 27 per-fold PROTAC
+synthesizability surrogate models (GNN:13, MLP:7, XGB:7), selected by greedy
+forward ensemble selection over 5x5-CV fold models -- see
+`scripts/models/evaluation.py`'s `EnsembleSelector.greedy_selection` /
+`run_ensemble_strategies` in the [RetroTAC](https://github.com/ribesstefano/RetroTAC) repo.
+
+- method: `caruana`
+- rmse: 0.1320
+
+## Usage
+
+```python
+from retrotac.models.ensemble import RetroTAC
+
+model = RetroTAC.from_pretrained("ailab-bio/RetroTAC")
+predictions = model.predict(["CCO", "c1ccccc1"])
+```

config.json

@@ -0,0 +1,9 @@
+{
+  "fp_size": 1024,
+  "fp_radius": 8,
+  "use_fingerprints": true,
+  "use_descriptors": true,
+  "target": "synthesizability",
+  "molecule_col": "smiles",
+  "source_config": "/proj/berzelius-2026-62/users/x_steri/RetroTAC/config/models_config.yaml"
+}

ensemble.json

@@ -0,0 +1,203 @@
+{
+  "created_at": "2026-09-02T16:26:13.437912",
+  "source_weights_file": "ensemble_weights_caruana.json",
+  "method": "caruana",
+  "metric_name": "rmse",
+  "metric_value": 0.1320195941716477,
+  "n_models": 27,
+  "members": {
+    "MLP_seed42_fold2": {
+      "backend": "mlp",
+      "seed": 42,
+      "fold": 2,
+      "weight": 0.20886679368103517,
+      "path": "members/mlp_seed42_fold2"
+    },
+    "GNN_seed123_fold3": {
+      "backend": "gnn",
+      "seed": 123,
+      "fold": 3,
+      "weight": 0.07996820349761527,
+      "path": "members/gnn_seed123_fold3"
+    },
+    "GNN_seed789_fold3": {
+      "backend": "gnn",
+      "seed": 789,
+      "fold": 3,
+      "weight": 0.07605156955002157,
+      "path": "members/gnn_seed789_fold3"
+    },
+    "XGB_seed789_fold2": {
+      "backend": "xgb",
+      "seed": 789,
+      "fold": 2,
+      "weight": 0.06511553051026735,
+      "path": "members/xgb_seed789_fold2"
+    },
+    "MLP_seed1011_fold3": {
+      "backend": "mlp",
+      "seed": 1011,
+      "fold": 3,
+      "weight": 0.06315293323807256,
+      "path": "members/mlp_seed1011_fold3"
+    },
+    "XGB_seed1011_fold4": {
+      "backend": "xgb",
+      "seed": 1011,
+      "fold": 4,
+      "weight": 0.05787625861155273,
+      "path": "members/xgb_seed1011_fold4"
+    },
+    "GNN_seed789_fold4": {
+      "backend": "gnn",
+      "seed": 789,
+      "fold": 4,
+      "weight": 0.051437246963562756,
+      "path": "members/gnn_seed789_fold4"
+    },
+    "MLP_seed456_fold0": {
+      "backend": "mlp",
+      "seed": 456,
+      "fold": 0,
+      "weight": 0.05135288002935062,
+      "path": "members/mlp_seed456_fold0"
+    },
+    "XGB_seed789_fold1": {
+      "backend": "xgb",
+      "seed": 789,
+      "fold": 1,
+      "weight": 0.05048265460030166,
+      "path": "members/xgb_seed789_fold1"
+    },
+    "GNN_seed789_fold1": {
+      "backend": "gnn",
+      "seed": 789,
+      "fold": 1,
+      "weight": 0.04841093117408907,
+      "path": "members/gnn_seed789_fold1"
+    },
+    "GNN_seed42_fold2": {
+      "backend": "gnn",
+      "seed": 42,
+      "fold": 2,
+      "weight": 0.035970171469397474,
+      "path": "members/gnn_seed42_fold2"
+    },
+    "GNN_seed789_fold0": {
+      "backend": "gnn",
+      "seed": 789,
+      "fold": 0,
+      "weight": 0.03523110661268556,
+      "path": "members/gnn_seed789_fold0"
+    },
+    "GNN_seed456_fold4": {
+      "backend": "gnn",
+      "seed": 456,
+      "fold": 4,
+      "weight": 0.027181546871949347,
+      "path": "members/gnn_seed456_fold4"
+    },
+    "GNN_seed1011_fold3": {
+      "backend": "gnn",
+      "seed": 1011,
+      "fold": 3,
+      "weight": 0.02714932126696833,
+      "path": "members/gnn_seed1011_fold3"
+    },
+    "XGB_seed42_fold2": {
+      "backend": "xgb",
+      "seed": 42,
+      "fold": 2,
+      "weight": 0.026671826625386995,
+      "path": "members/xgb_seed42_fold2"
+    },
+    "GNN_seed42_fold1": {
+      "backend": "gnn",
+      "seed": 42,
+      "fold": 1,
+      "weight": 0.016036036036036035,
+      "path": "members/gnn_seed42_fold1"
+    },
+    "GNN_seed123_fold2": {
+      "backend": "gnn",
+      "seed": 123,
+      "fold": 2,
+      "weight": 0.013229018492176387,
+      "path": "members/gnn_seed123_fold2"
+    },
+    "XGB_seed1011_fold3": {
+      "backend": "xgb",
+      "seed": 1011,
+      "fold": 3,
+      "weight": 0.011655405405405406,
+      "path": "members/xgb_seed1011_fold3"
+    },
+    "GNN_seed456_fold0": {
+      "backend": "gnn",
+      "seed": 456,
+      "fold": 0,
+      "weight": 0.011513157894736841,
+      "path": "members/gnn_seed456_fold0"
+    },
+    "GNN_seed42_fold3": {
+      "backend": "gnn",
+      "seed": 42,
+      "fold": 3,
+      "weight": 0.008333333333333333,
+      "path": "members/gnn_seed42_fold3"
+    },
+    "GNN_seed1011_fold0": {
+      "backend": "gnn",
+      "seed": 1011,
+      "fold": 0,
+      "weight": 0.00625,
+      "path": "members/gnn_seed1011_fold0"
+    },
+    "MLP_seed456_fold1": {
+      "backend": "mlp",
+      "seed": 456,
+      "fold": 1,
+      "weight": 0.00625,
+      "path": "members/mlp_seed456_fold1"
+    },
+    "MLP_seed1011_fold2": {
+      "backend": "mlp",
+      "seed": 1011,
+      "fold": 2,
+      "weight": 0.0058823529411764705,
+      "path": "members/mlp_seed1011_fold2"
+    },
+    "XGB_seed42_fold4": {
+      "backend": "xgb",
+      "seed": 42,
+      "fold": 4,
+      "weight": 0.005263157894736842,
+      "path": "members/xgb_seed42_fold4"
+    },
+    "XGB_seed456_fold4": {
+      "backend": "xgb",
+      "seed": 456,
+      "fold": 4,
+      "weight": 0.005263157894736842,
+      "path": "members/xgb_seed456_fold4"
+    },
+    "MLP_seed456_fold2": {
+      "backend": "mlp",
+      "seed": 456,
+      "fold": 2,
+      "weight": 0.002702702702702703,
+      "path": "members/mlp_seed456_fold2"
+    },
+    "MLP_seed1011_fold4": {
+      "backend": "mlp",
+      "seed": 1011,
+      "fold": 4,
+      "weight": 0.002702702702702703,
+      "path": "members/mlp_seed1011_fold4"
+    }
+  },
+  "metadata": {
+    "n_selection_samples": 463,
+    "n_evaluation_samples": 1854
+  }
+}

members/gnn_seed1011_fold0.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e19943524b062d9efcc4d8f66d7892a8957d22209a218e6758d893c0c3ae7b36
+size 107739362

members/gnn_seed1011_fold3.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2e470a30034f8c918f4b5e35a5e353425d0fa0ef8661bcfa540faaad0d0e4038
+size 107739362

members/gnn_seed123_fold2.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fa8a5d8c47f5ad22200b4fc9a6bd4cad2d9b80472a9127bf636a7b03c2cd86b
+size 120339850

members/gnn_seed123_fold3.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:760810070f05827527d36c9217eab3f76d79e08e42ebeb02ac61b87ad31597d4
+size 110888162

members/gnn_seed42_fold1.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4ea85c6ea2e71705a8afb95a52199d1bc694533b3b5f2a17638aeeb4902e8a1e
+size 111679882

members/gnn_seed42_fold2.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c2763e3c30966161cadc3236b25cddd70fd1e62cb46c4b1ec741ded82f843c2f
+size 112471794

members/gnn_seed42_fold3.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f0ad509ae1107bc6bc43b456f20444b8ac7a5f5a6bb56e5d2062593bacf0ebe
+size 107939722

members/gnn_seed456_fold0.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a14d3c6be00b1949aceccaa84eebe217d36a7e44c6fbad4d12855bc1f35c1310
+size 107939722

members/gnn_seed456_fold4.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:edf99fa93ecb887e5d24acb8c165c5a9c05db66b33be0a48054ae69e30c438d9
+size 108140210

members/gnn_seed789_fold0.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aebc81226f36fd8b94622ffa015a34b516f75c844a75e38ee42fee7385dac9a9
+size 112471794

members/gnn_seed789_fold1.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:29d1d36ecd1f75b3f7aee7175cf121c523d1a3c0a970c9bef000416547788eee
+size 120339850

members/gnn_seed789_fold3.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7e44a1092da252024f314bcac56d9e72b4d317d5f826bd76f6efcbd1c2ecc8c6
+size 117185762

members/gnn_seed789_fold4.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:023fb5dfc4426a52979d65d897e0e8c37bf80df5054b674bccf80c866ef92741
+size 107739362

members/mlp_seed1011_fold2.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:abe4ab791f1af918eea4517c605fd81cd65deb5fe5483f5aa251d8badb511213
+size 124887

members/mlp_seed1011_fold2.skops

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e074dc63bce127293f420f55727096aebe868db17df55e6494614476a9e544cd
+size 1149643

members/mlp_seed1011_fold3.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5677794649859855379fa4ed7e9fd0f6fd3a1784b467fa670d3d4badb60c029c
+size 317836

members/mlp_seed1011_fold3.skops

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:31410e14dd79a3c4ec930d1c70ff2f12c90b16d6caf7222a71c45cf0375949be
+size 1104481

members/mlp_seed1011_fold4.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:85c475c73662165965bdac44d1e33c6125664a6369a5bb2e4de8725a500357d0
+size 224657

members/mlp_seed1011_fold4.skops

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:598f652e819100acdf0fb2c09c012cea0dbe7628a45511c0bec9dcfbb6665d0c
+size 1059238

members/mlp_seed42_fold2.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1ffe0adc4c2e31ce0a28d28e29f7c88dcabf2632c9a33f152cee46886a42e259
+size 107933

members/mlp_seed42_fold2.skops

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cf68df05a3da17ec208001124e35e11cd1c78dde3141cad1158fef606fd9277d
+size 878502

members/mlp_seed456_fold0.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dd3bb82762ab247ca5e7e23cf3e472896fa32fc5c07964166de595e167d54446
+size 1057657

members/mlp_seed456_fold0.skops

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7eb7750eda999dc8def5048175f55819458fb2746e2179bd2d622ea31fa5724e
+size 829238

members/mlp_seed456_fold1.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b4fe42889c03201493a207270f932d49c19d6982f2f621f152c54214f3f63824
+size 1051513

members/mlp_seed456_fold1.skops

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:34fcf9b6917bbc32c15a72bb4b6c8d89fb84240c430fe601343f97f5079858e1
+size 804584

members/mlp_seed456_fold2.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f25634aec7a62d24151b485b6f34adefbaff3529e8ed70f7f41ff0d9954214da
+size 361273

members/mlp_seed456_fold2.skops

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a1d14198dc07f8ca24a9c1c66fb0e501bc26f185d90f6db63c404b5f216a76fa
+size 1026376

members/xgb_seed1011_fold3.skops

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d5942a9631c31675150687d08875c11d5dd7efa61d7dacc2f1c50ec6e27d7cba
+size 944593

members/xgb_seed1011_fold3.ubj

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:569b1b0050b76984f4b517caf18a2c78bdd619faa1a5e3c0a8324c3a236d6a06
+size 4796184

members/xgb_seed1011_fold4.skops

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8842399a28146e1bd15f20086b088ad5a1d596d8616972aa8e458e8037732e34
+size 915789

members/xgb_seed1011_fold4.ubj

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e127501c116bac1bf4b82509048f2744866b9553b89b2ea250c7f15d1c439133
+size 3071716

members/xgb_seed42_fold2.skops

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:423d044ae6d5cc3cd2e7bfb6344b81efb50db0dd01e81021b3ec3e1cc17ea9fe
+size 1145868

members/xgb_seed42_fold2.ubj

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b50ec15fdc39a7bf879205f5eae6ebf59aa55d5db6b5f620c0f1d0eba8d91d47
+size 7271863

members/xgb_seed42_fold4.skops

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:918a674c08e9f20f9632ed76908f1f5a6703c3bedab418a78487e71fa723b560
+size 1186944

members/xgb_seed42_fold4.ubj

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:97f42e4644ed8a65d591c7afde6f7d88787966b18c9908d32d095b538f60af32
+size 2939538

members/xgb_seed456_fold4.skops

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6a2d4bdd446864b67a92e2b4a8629a2b8e8b801074dde1b97d9225cf3081f749
+size 1145841

members/xgb_seed456_fold4.ubj

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6d6d812da629a6b9fbdffb03261127802a0bb3d88f5ba462c57f613f7c768cb5
+size 6575979

members/xgb_seed789_fold1.skops

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fb059aadcfc3247b99a413a26abb01b17dc0d43c8c7692e4b154077ff7549246
+size 792573

members/xgb_seed789_fold1.ubj

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:800463c8b8cb59457dc29b871512c2738571c3e157360dae1b2be8d3074a1464
+size 2911894

members/xgb_seed789_fold2.skops

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:117dc3c5b179195e2e0b7a58e1b3c86d6e38a07e29a597ab06c4b185e91230fc
+size 924029

members/xgb_seed789_fold2.ubj

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72eac8b949c9c5068906b2d4e402d49cc575cfbd4262bc0ccddd488526f8a86c
+size 7412111