Upload BBB classifier (exp06_esm_tab_mlp_aug)

README.md

@@ -0,0 +1,63 @@
+---
+license: mit
+tags:
+  - peptide
+  - blood-brain-barrier
+  - bbb
+  - classification
+  - esm
+library_name: bbb-models
+pipeline_tag: text-classification
+
+base_model: manumartinm/bbb-classifier
+---
+
+# BBB Classifier — `exp06_esm_tab_mlp_aug`
+
+Sequence/tabular BBB permeability classifier for peptide candidates. Fuses ESM-2 embeddings with physicochemical descriptors and isotonic-calibrated probabilities.
+
+## Model summary
+
+| Field | Value |
+|-------|-------|
+| Experiment | `exp06_esm_tab_mlp_aug` |
+| Architecture | `esm_tab_mlp` |
+| ESM dim | `128` |
+| Hidden dim | `256` |
+| PR-AUC (calibrated) | `0.906775452423001` |
+| ROC-AUC (calibrated) | `0.9185328185328185` |
+| MCC (calibrated) | `0.6435046074523986` |
+
+## Files
+
+- `checkpoints/best.ckpt` — PyTorch weights
+- `calibrators/calibrator.pkl` — isotonic calibration
+- `train_metadata.json` — experiment + data config for inference
+- `metrics.json` — validation metrics
+
+## Usage
+
+Install the project and download this repo:
+
+```bash
+git clone https://github.com/your-org/TFG.git
+cd TFG/bbb_models && uv sync
+hf download manumartinm/bbb-classifier --local-dir ./bbb-classifier
+```
+
+Score candidates (input must include sequence + tabular descriptor columns used at training time):
+
+```bash
+uv run python scripts/classifier/predict.py \
+  --run-dir ./bbb-classifier \
+  --input candidates.parquet \
+  --output scored.parquet
+```
+
+## Dataset
+
+Trained on the TFG BBB peptide dataset. Companion dataset: [`manumartinm/bbb-peptides`](https://huggingface.co/datasets/manumartinm/bbb-peptides).
+
+## Citation
+
+If you use this model, cite the TFG BBB peptide modeling pipeline.

calibrators/calibrator.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2a59f4633d4aa2d83a1dc5d5400b13323a1ee24e3981bd4aacc4b98964b4ba62
+size 704

checkpoints/best.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4a1dce08cc4782397df42a192527a7b33a906c0164416c3880e335518ab9e185
+size 1838017

export_stats.json

@@ -0,0 +1,15 @@
+{
+  "kind": "classifier",
+  "run_dir": "/Users/manumartinm/Documents/ProteinDesign/TFG/bbb_models/artifacts/models/exp06_esm_tab_mlp_aug",
+  "output_dir": "/Users/manumartinm/Documents/ProteinDesign/TFG/bbb_models/artifacts/hf_release/bbb-classifier",
+  "run_name": "exp06_esm_tab_mlp_aug",
+  "model_type": "esm_tab_mlp",
+  "files": [
+    "checkpoints/best.ckpt",
+    "calibrators/calibrator.pkl",
+    "train_metadata.json",
+    "metrics.json",
+    "README.md"
+  ],
+  "repo_id": "manumartinm/bbb-classifier"
+}

metrics.json

@@ -0,0 +1,22 @@
+{
+  "calibrated": {
+    "brier": 0.1134513631011226,
+    "ece": 5.7948960259975495e-09,
+    "mcc": 0.6435046074523986,
+    "pr_auc": 0.906775452423001,
+    "precision": 0.8666666666666667,
+    "roc_auc": 0.9185328185328185,
+    "sensitivity": 0.7428571428571429
+  },
+  "model_type": "esm_tab_mlp",
+  "raw": {
+    "brier": 0.13664215418462355,
+    "ece": 0.06837135491271813,
+    "mcc": 0.6435046074523986,
+    "pr_auc": 0.9094165678023901,
+    "precision": 0.8666666666666667,
+    "roc_auc": 0.8965250965250965,
+    "sensitivity": 0.7428571428571429
+  },
+  "run_name": "exp06_esm_tab_mlp_aug"
+}

train_metadata.json

@@ -0,0 +1,115 @@
+{
+  "data_cfg": {
+    "dataset_path": "../dataset/data/processed/peptides_bbb_with_augmentation.parquet",
+    "fold_col": "fold_id",
+    "id_col": "peptide_id",
+    "label_col": "bbb_label",
+    "random_state": 42,
+    "sequence_col": "sequence",
+    "struct_manifest_path": "../dataset/data/processed/peptides_struct_manifest.parquet",
+    "tabular_exclude": [
+      "peptide_id",
+      "sequence",
+      "bbb_label",
+      "split",
+      "source_split",
+      "source_db",
+      "label_tier",
+      "is_gold",
+      "cluster_id",
+      "external_test",
+      "source_id",
+      "assay_method",
+      "reference",
+      "organism",
+      "fold_id",
+      "is_augmented",
+      "parent_peptide_id",
+      "sample_weight"
+    ],
+    "test_size": 0.2,
+    "three_d_columns": []
+  },
+  "exp_cfg": {
+    "esm": {
+      "cache_dir": "artifacts/cache/esm2"
+    },
+    "features": {
+      "use_3d": false,
+      "use_esm": true,
+      "use_gnn": false,
+      "use_tabular": true
+    },
+    "mixup": {
+      "alpha": 0.2,
+      "enabled": true,
+      "prob": 0.5
+    },
+    "model": {
+      "dropout": 0.2,
+      "esm_dim": 128,
+      "hidden_dim": 256
+    },
+    "model_type": "esm_tab_mlp",
+    "name": "exp06_esm_tab_mlp_aug"
+  },
+  "tab_cols": [
+    "length",
+    "mw",
+    "ext_coef_reduced",
+    "ext_coef_oxidized",
+    "hydrophobic_ratio_pct",
+    "pi",
+    "net_charge_ph7",
+    "total_charge",
+    "mean_hydrophobicity",
+    "hydrophobicity_ph7",
+    "hydrophilic_ratio",
+    "aliphatic_index",
+    "boman_index",
+    "aromaticity",
+    "instability_index",
+    "gravy",
+    "charge_density",
+    "aa_basic_pct",
+    "aa_acidic_pct",
+    "aa_aromatic_pct",
+    "aa_hydrophobic_pct",
+    "aa_polar_pct",
+    "hydrophobic_moment",
+    "mw_pyteomics",
+    "mw_delta_abs"
+  ],
+  "train_cfg": {
+    "calibration": {
+      "enabled": true,
+      "method": "isotonic"
+    },
+    "maximize_metric": true,
+    "output": {
+      "keep_top_k": 1,
+      "root": "artifacts",
+      "save_periodic_every": 5
+    },
+    "primary_metric": "pr_auc",
+    "run_name": "default",
+    "secondary_metric": "mcc",
+    "seed": 42,
+    "tracking": {
+      "mlflow": true,
+      "mlflow_experiment": "bbb_classifier",
+      "tensorboard": true
+    },
+    "training": {
+      "batch_size": 128,
+      "epochs": 50,
+      "eval_every": 1,
+      "grad_clip": 1.0,
+      "log_every": 5,
+      "lr": 0.001,
+      "num_workers": 4,
+      "patience": 8,
+      "weight_decay": 0.0001
+    }
+  }
+}