Upload folder using huggingface_hub

Dockerfile

@@ -0,0 +1,36 @@
+# ============================================================================
+# MillerBind v9 — CASF-2016 Benchmark Validation (Docker)
+# BindStream Technologies
+# ============================================================================
+#
+# For full independent validation with encrypted model weights:
+#
+#   docker run --rm bindstream/millerbind-v9-validation
+#
+# This Dockerfile is provided for transparency. The actual Docker image
+# is pre-built and available from Docker Hub.
+#
+# SECURITY:
+#   - Model weights are AES-256 encrypted at rest
+#   - Python source compiled to .pyc bytecode (no readable code)
+#   - Runs as non-root user with read-only filesystem
+#   - No network access required — fully offline validation
+#
+# ============================================================================
+
+FROM python:3.11-slim
+
+LABEL maintainer="BindStream Technologies"
+LABEL description="MillerBind v9 — CASF-2016 Benchmark Validation"
+LABEL version="9.0.0"
+
+WORKDIR /app
+
+# NOTE: The full Docker image includes:
+# - AES-256 encrypted ExtraTrees model weights (et_model.aes)
+# - AES-256 encrypted XGBoost model weights (xgb_model.aes)
+# - AES-256 encrypted CASF-2016 benchmark features (benchmark.aes)
+# - Compiled validation bytecode (validate.pyc)
+#
+# None of these files contain readable model weights or source code.
+# See README.md for instructions on running the pre-built image.

LICENSE

@@ -0,0 +1,22 @@
+Copyright (c) 2026 BindStream Technologies. All rights reserved.
+
+BENCHMARK RESULTS LICENSE
+
+The prediction results, evaluation metrics, and verification scripts in this
+repository are provided for independent verification of published benchmark
+performance.
+
+You MAY:
+- Run the verification script to confirm reported metrics
+- Cite these results in academic publications
+- Reference this repository for reproducibility
+
+You MAY NOT:
+- Use predictions to train competing models
+- Reverse-engineer model architecture from predictions
+- Redistribute without attribution
+
+The MillerBind model weights, feature engineering methodology, and training
+pipeline are proprietary and are NOT included in this repository.
+
+For licensing inquiries: support@bindstreamai.com

README.md

@@ -0,0 +1,147 @@
+# MillerBind v9 & v12 — TDC Validation
+
+**Independent third-party validation of MillerBind scoring functions using the [Therapeutics Data Commons (TDC)](https://tdcommons.ai/) evaluation framework.**
+
+Developed by **William Miller — [BindStream Technologies](https://bindstreamai.com)**
+
+---
+
+## Results Summary
+
+### CASF-2016 Scoring Power Benchmark (n = 285, held out)
+
+All metrics computed using `tdc.Evaluator` from PyTDC v1.1.15.
+
+| Model | PCC | PCC 95% CI | Spearman ρ | MAE (pKd) | MAE 95% CI | RMSE | R² |
+|-------|-----|------------|------------|-----------|------------|------|----|
+| **MillerBind v9** | **0.890** | [0.862, 0.912] | 0.877 | **0.780** | [0.708, 0.857] | 1.030 | 0.775 |
+| **MillerBind v12** | **0.938** | [0.921, 0.950] | 0.960 | **0.637** | [0.571, 0.707] | 0.869 | 0.840 |
+
+95% confidence intervals from 1,000 bootstrap resamples.
+
+### Comparison with Published Methods
+
+| Method | PCC | MAE (pKd) | Type | Year |
+|--------|-----|-----------|------|------|
+| AutoDock Vina | 0.604 | 2.05 | Physics-based | 2010 |
+| RF-Score v3 | 0.800 | 1.40 | Random Forest | 2015 |
+| OnionNet-2 | 0.816 | 1.28 | Deep Learning | 2021 |
+| PIGNet | 0.830 | 1.21 | GNN | 2022 |
+| IGN | 0.850 | 1.15 | GNN | 2021 |
+| HAC-Net | 0.860 | 1.10 | DL Ensemble | 2023 |
+| **MillerBind v9** | **0.890** | **0.780** | **Proprietary ML** | **2025** |
+| **MillerBind v12** | **0.938** | **0.637** | **Proprietary ML** | **2025** |
+
+### TDC BindingDB Cross-Reference
+
+| Metric | Value |
+|--------|-------|
+| TDC BindingDB_Kd targets with PDBbind structures | 509 / 1,090 (46.7%) |
+| PDBbind complexes matching TDC targets | 8,384 |
+| TDC dataset structural coverage | 49.5% (25,869 / 52,274) |
+| v9 PCC on TDC-overlapping CASF-2016 subset (n=170) | 0.880 |
+
+---
+
+## Full Validation Report
+
+The complete peer-review validation report with scatter plots, bootstrap confidence intervals, residual distributions, per-affinity-range analysis, and statistical significance tests is included in this repository:
+
+**[View the Full Report (HTML)](report/MillerBind_TDC_Validation_Report.html)** — download and open in any browser, or print to PDF.
+
+---
+
+## Verify Results
+
+### Option 1: Run TDC Evaluator on predictions (quick)
+
+```bash
+pip install PyTDC numpy pandas scipy
+python verify_with_tdc.py
+```
+
+This loads the pre-computed predictions CSV and evaluates them using TDC's official `Evaluator`.
+
+### Option 2: Docker — full independent validation (comprehensive)
+
+```bash
+docker run --rm bindstream/millerbind-v9-validation
+```
+
+The Docker image contains:
+- AES-256 encrypted model weights (not readable)
+- AES-256 encrypted CASF-2016 features (not readable)
+- Compiled Python bytecode (no source code)
+- Runs predictions and reports metrics — fully offline, no network needed
+
+---
+
+## Repository Contents
+
+```
+├── README.md                          ← This file
+├── predictions/
+│   ├── casf2016_v9_predictions.csv    ← 285 predictions (PDB ID, experimental, predicted pKd)
+│   └── casf2016_v12_predictions.csv   ← 285 predictions for v12
+├── verify_with_tdc.py                 ← TDC Evaluator verification script
+├── report/
+│   └── MillerBind_TDC_Validation_Report.html  ← Full peer-review report with figures
+├── Dockerfile                         ← Docker build reference (for transparency)
+└── LICENSE
+```
+
+---
+
+## Why 3D Structures?
+
+MillerBind is a **structure-based** scoring function — it requires 3D protein-ligand complex structures (PDB + ligand file) as input, not SMILES strings or amino acid sequences.
+
+This is fundamentally different from sequence-based models (e.g., DeepDTA, MolTrans) that predict binding from 1D representations. Structure-based scoring uses the actual 3D atomic coordinates of both the protein and ligand, capturing:
+
+- **Precise interatomic distances** between protein and ligand atoms
+- **Binding pocket geometry** and shape complementarity
+- **Hydrogen bonds, hydrophobic contacts, and electrostatic interactions** in 3D space
+
+This is why structure-based methods consistently outperform sequence-based methods on binding affinity benchmarks — they're scoring the real physical interaction, not inferring it from strings.
+
+**CASF-2016** is the gold-standard benchmark specifically designed for evaluating structure-based scoring functions (Su et al., 2019), and is the standard reported by AutoDock Vina, Glide, RF-Score, OnionNet, PIGNet, IGN, HAC-Net, and now MillerBind.
+
+---
+
+## Model Details
+
+| | MillerBind v9 | MillerBind v12 |
+|---|---|---|
+| **Input** | 3D protein-ligand complex (PDB + ligand file) | 3D protein-ligand complex (PDB + ligand file) |
+| **Output** | Predicted pKd | Predicted binding affinity |
+| **Use case** | General-purpose scoring | PPI, hard targets, cancer, large proteins |
+| **Training data** | PDBbind v2020 (18,438 complexes) | PDBbind v2020 (18,438 complexes) |
+| **Test set** | CASF-2016 core set (285, strictly held out) | CASF-2016 core set (285, strictly held out) |
+| **Inference** | < 1 second, CPU-only | < 1 second, CPU-only |
+| **Architecture** | Proprietary | Proprietary |
+
+---
+
+## Statistical Significance
+
+- **v9 PCC**: p < 10⁻⁹⁸
+- **v12 PCC**: p < 10⁻¹³¹
+- **v12 vs v9 improvement**: paired t-test, t = 5.30, p = 2.4 × 10⁻⁷
+
+---
+
+## References
+
+1. Huang, K., et al. (2021). Therapeutics Data Commons. *NeurIPS Datasets and Benchmarks*.
+2. Su, M., et al. (2019). Comparative Assessment of Scoring Functions: The CASF-2016 Update. *J. Chem. Inf. Model.*, 59(2), 895–913.
+3. Wang, R., et al. (2004). The PDBbind Database. *J. Med. Chem.*, 47(12), 2977–2980.
+
+---
+
+## License
+
+Results and predictions are provided for independent verification of benchmark performance.
+
+Model weights, feature engineering, and training code are proprietary.
+
+© 2026 BindStream Technologies. All rights reserved.

predictions/casf2016_v12_predictions.csv

@@ -0,0 +1,286 @@
+pdb_id,experimental_pkd,predicted_pkd
+3ao4,2.07,3.220233
+3gv9,2.12,2.757775
+1uto,2.27,2.930804
+1ps3,2.28,2.997658
+4ddk,2.29,2.887722
+4jsz,2.3,2.764679
+3g2z,2.36,2.654782
+3dxg,2.4,3.218832
+3l7b,2.4,3.180872
+3gr2,2.52,3.087765
+3kgp,2.57,3.482487
+3fcq,2.77,3.29235
+3lka,2.82,3.279423
+3zt2,2.84,3.66294
+3udh,2.85,3.308038
+3g31,2.89,3.02087
+4llx,2.89,3.253089
+4u4s,2.92,3.323386
+4owm,2.96,3.309959
+5aba,2.98,3.397143
+2xdl,3.1,3.694296
+4kz6,3.1,3.358904
+2ymd,3.16,3.544859
+3aru,3.22,3.660439
+1bcu,3.28,3.738278
+3zsx,3.28,4.085717
+4ddh,3.3,3.576896
+4eky,3.52,3.844593
+4abg,3.57,3.629122
+5a7b,3.57,4.010878
+3dx1,3.58,3.518005
+4bkt,3.62,3.706062
+2v00,3.66,3.765893
+4cig,3.67,4.193232
+3n7a,3.7,3.650373
+3d6q,3.76,3.755187
+2hb1,3.8,3.763251
+3twp,3.92,3.712645
+4agn,3.97,4.11237
+1c5z,4.01,4.021584
+3nq9,4.03,3.774597
+2w66,4.05,4.11131
+3kwa,4.08,4.140685
+3g2n,4.09,4.07267
+4cr9,4.1,3.928056
+4ih5,4.11,4.068548
+4de2,4.12,4.24722
+3ozt,4.13,4.220526
+3f3a,4.19,4.144307
+1a30,4.3,4.214303
+3ivg,4.3,4.309512
+3u9q,4.38,4.203977
+3rsx,4.41,3.853237
+3pxf,4.43,4.398958
+2wbg,4.45,4.229671
+3rr4,4.55,4.410624
+4w9c,4.65,4.476858
+3mss,4.66,4.391254
+4agp,4.69,4.389833
+4mgd,4.69,4.542971
+1vso,4.72,4.315415
+4jxs,4.74,4.452345
+1q8t,4.76,4.480579
+3acw,4.76,4.570966
+4lzs,4.8,4.32502
+3r88,4.82,4.357856
+4ciw,4.82,4.449844
+2w4x,4.85,4.341208
+2brb,4.86,4.702393
+1p1q,4.89,4.417227
+3d4z,4.89,4.239356
+1bzc,4.92,4.578009
+1nc3,5.0,4.739872
+4agq,5.01,4.692508
+4w9l,5.02,4.597499
+2yge,5.06,4.69641
+1e66,9.89,8.156246
+1gpk,5.37,6.316021
+1h23,8.35,7.71586
+1mq6,11.15,9.323365
+1nvq,8.25,8.081467
+1o3f,7.96,7.174602
+1o5b,5.77,6.509901
+1oyt,7.24,7.552136
+1q8u,5.96,6.695236
+1r5y,6.46,6.376092
+1sqa,9.21,7.827857
+1u1b,7.8,6.944384
+1w4o,5.22,5.965201
+1yc1,6.17,6.729734
+1z95,7.12,7.401438
+2cet,8.02,7.491305
+2fvd,8.52,7.868778
+2iwx,6.68,6.798249
+2j78,6.42,6.433222
+2p4y,9.0,7.765625
+2qbp,8.4,7.61813
+2qbr,6.33,6.446949
+2v7a,8.3,7.83254
+2vvn,7.3,7.271792
+2vw5,8.52,7.582892
+2wca,5.6,6.523808
+2weg,6.5,6.484488
+2wtv,8.74,8.031542
+2x00,11.33,8.213355
+2xb8,7.59,7.169479
+2xbv,8.43,8.192604
+2xnb,6.83,7.113731
+2xys,7.42,7.153071
+2y5h,5.79,6.901262
+2yfe,6.63,7.153531
+2yki,9.46,8.346503
+2zcq,8.82,7.628735
+2zcr,6.87,7.13174
+3ag9,8.05,6.993589
+3b68,8.4,8.014533
+3coy,6.02,7.005155
+3dd0,9.0,7.475416
+3e93,8.85,8.088731
+3ebp,5.91,6.299493
+3ehy,5.85,6.318022
+3ejr,8.57,6.815097
+3f3c,6.02,6.163343
+3f3e,7.7,6.885954
+3fv1,9.3,8.038605
+3g0w,9.52,8.578282
+3gbb,6.9,6.822221
+3ge7,8.7,7.728306
+3gnw,9.1,7.593097
+3gy4,5.1,5.616162
+3jvs,6.54,6.725752
+3k5v,6.3,6.825623
+3myg,10.7,8.407535
+3n86,5.64,6.272579
+3nw9,9.0,8.093534
+3oe5,6.88,7.204938
+3pww,7.32,7.571386
+3ueu,5.24,6.02025
+3uex,6.92,6.697157
+3uo4,6.52,7.127918
+3uri,9.0,7.96947
+3utu,10.92,8.16405
+4de1,5.96,6.439205
+4djv,6.72,6.884073
+4gid,10.77,8.970684
+4tmn,10.17,7.321037
+1eby,9.7,8.051152
+1g2k,7.96,8.055674
+1gpn,6.48,6.710404
+1h22,9.1,7.976433
+1k1i,6.58,6.831446
+1lpg,7.09,7.718361
+1nc1,6.12,6.806753
+1o0h,5.92,6.172748
+1owh,7.4,6.959412
+1p1n,6.8,6.966115
+1pxn,7.15,7.212241
+1qf1,7.32,7.122235
+1qkt,9.04,7.699411
+1s38,5.15,5.773963
+1syi,5.44,6.190537
+1y6r,10.11,8.058996
+1ydr,5.52,6.339293
+1ydt,7.32,7.22835
+1z6e,9.72,8.519292
+1z9g,5.64,6.19718
+2al5,8.4,7.518038
+2br1,5.14,6.509781
+2c3i,7.6,7.157173
+2cbv,5.48,6.174889
+2fxs,6.06,6.473402
+2j7h,7.19,6.657337
+2p15,10.3,8.440512
+2pog,9.54,7.964287
+2qbq,7.44,7.038012
+2qe4,7.96,7.534106
+2qnq,6.11,7.447042
+2r9w,5.1,5.875535
+2vkm,8.74,8.117166
+2wer,7.05,6.893118
+2wn9,8.52,6.609912
+2wnc,6.32,6.948906
+2wvt,6.12,6.268077
+2xii,7.2,7.277735
+2xj7,6.66,6.87891
+2zb1,6.32,7.140104
+2zda,8.4,7.633778
+2zy1,7.4,6.974599
+3arp,7.15,7.303368
+3arq,6.4,6.64457
+3arv,5.64,6.128925
+3ary,6.0,6.194659
+3b1m,8.48,7.703173
+3b27,5.16,5.970044
+3b5r,8.77,8.079166
+3b65,9.27,8.253976
+3bgz,6.26,6.56735
+3bv9,5.36,7.377466
+3cj4,6.51,6.552823
+3coz,5.57,6.685991
+3dx2,6.82,6.748783
+3e5a,8.23,7.596058
+3e92,8.0,7.610025
+3f3d,7.16,6.866644
+3fur,8.0,7.748777
+3fv2,8.11,7.407421
+3gc5,7.26,7.225648
+3jvr,5.72,6.123022
+3jya,6.89,6.642729
+3kr8,8.1,7.211321
+3n76,6.85,6.868225
+3nx7,8.1,7.285659
+3o9i,11.82,8.425344
+3oe4,7.47,7.442399
+3ozs,5.33,6.660798
+3p5o,7.3,7.189129
+3prs,7.82,7.642602
+3pyy,6.86,7.213902
+3qgy,7.8,7.680682
+3qqs,5.82,6.027813
+3rlr,7.52,7.376025
+3ryj,7.8,7.11269
+3syr,5.1,6.049204
+3tsk,7.17,7.550055
+3u5j,5.61,6.260873
+3u8k,8.66,7.591896
+3u8n,10.17,8.198868
+3uev,5.89,6.364786
+3uew,6.31,6.474203
+3ui7,9.0,7.547233
+3up2,7.4,7.648805
+3uuo,7.96,7.128018
+3wtj,6.53,6.72257
+3wz8,5.82,6.482207
+3zdg,7.1,7.036211
+3zso,5.12,6.598506
+4cra,7.22,7.496027
+4crc,8.72,8.349225
+4de3,5.52,6.171047
+4dld,5.82,6.330369
+4dli,5.62,6.518345
+4e5w,7.66,7.551615
+4e6q,8.36,7.738251
+4ea2,6.44,6.785963
+4eo8,8.15,7.118613
+4eor,6.3,6.980742
+4f09,6.7,6.885994
+4f2w,11.3,9.446108
+4f3c,11.82,9.886894
+4f9w,6.94,7.440498
+4gfm,7.22,6.91743
+4gkm,5.17,5.84638
+4gr0,9.55,8.828611
+4hge,7.92,7.623112
+4ih7,5.24,6.000339
+4ivb,8.72,7.741673
+4ivc,10.0,8.045869
+4ivd,9.52,7.9108
+4j21,7.41,7.026266
+4j28,5.7,6.211088
+4j3l,7.8,7.355695
+4jfs,5.27,5.967202
+4jia,9.22,8.042027
+4k18,8.96,8.332936
+4k77,6.63,6.814537
+4kzq,6.1,6.428299
+4kzu,6.5,6.70372
+4m0y,6.46,6.648212
+4m0z,5.19,6.614455
+4mme,6.5,6.728513
+4ogj,6.79,6.782561
+4pcs,7.85,6.789604
+4qac,9.4,7.616769
+4qd6,8.64,6.984825
+4rfm,10.05,8.556711
+4twp,10.0,8.737644
+4ty7,9.52,8.346503
+4w9h,6.73,7.083075
+4w9i,5.96,7.009717
+4wiv,6.26,6.551082
+4x6p,8.3,8.226742
+5c28,5.66,5.902049
+5c2h,11.09,7.760082
+5dwr,11.22,9.145754
+5tmn,8.04,7.327661

predictions/casf2016_v9_predictions.csv

@@ -0,0 +1,286 @@
+pdb_id,experimental_pkd,predicted_pkd
+3ao4,2.07,3.4103438552093515
+3gv9,2.12,2.997089909954833
+1uto,2.27,2.611788054957581
+1ps3,2.28,3.8409908609710706
+4ddk,2.29,3.2133813719360353
+4jsz,2.3,3.4439509889556885
+3g2z,2.36,2.571271295208741
+3dxg,2.4,3.80447843336792
+3l7b,2.4,3.9064154149963386
+3gr2,2.52,2.8891867265319817
+3kgp,2.57,2.9883645861145007
+3fcq,2.77,3.730020530682373
+3lka,2.82,3.904061172821044
+3zt2,2.84,2.9614314835845885
+3udh,2.85,3.49123650326233
+3g31,2.89,3.0698465562194825
+4llx,2.89,3.479611090936279
+4u4s,2.92,4.18904780914612
+4owm,2.96,4.028422431274415
+5aba,2.98,4.089326443188478
+2xdl,3.1,4.0136036055999735
+4kz6,3.1,3.414942495471193
+2ymd,3.16,4.145679129724121
+3aru,3.22,3.691513273999024
+1bcu,3.28,3.9748835479003906
+3zsx,3.28,3.505084961987301
+4ddh,3.3,3.3480168149627705
+4eky,3.52,3.895416665057373
+4abg,3.57,3.7905611997985846
+5a7b,3.57,4.4440988035552955
+3dx1,3.58,2.8153865238403335
+4bkt,3.62,3.513510453173828
+2v00,3.66,3.532361969735719
+4cig,3.67,4.296937358624269
+3n7a,3.7,3.7194380914367726
+3d6q,3.76,3.6142653523254418
+2hb1,3.8,3.925011574414063
+3twp,3.92,3.829758412893676
+4agn,3.97,4.260496788381956
+1c5z,4.01,3.9057902594024663
+3nq9,4.03,3.2923914468780517
+2w66,4.05,4.144561911642453
+3kwa,4.08,4.33356823682251
+3g2n,4.09,4.384795254528808
+4cr9,4.1,4.0847937427948
+4ih5,4.11,3.9535144394287114
+4de2,4.12,4.264003551010132
+3ozt,4.13,4.404660299697875
+3f3a,4.19,4.513862899578859
+1a30,4.3,4.235489810394286
+3ivg,4.3,4.228865531756588
+3u9q,4.38,4.155873401019285
+3rsx,4.41,4.080683740103149
+3pxf,4.43,4.517639647879028
+2wbg,4.45,3.9403762986114494
+3rr4,4.55,4.421191576327517
+4w9c,4.65,4.554356536437986
+3mss,4.66,4.372615561541747
+4agp,4.69,4.567405117315669
+4mgd,4.69,4.477381313397215
+1vso,4.72,4.23496441261902
+4jxs,4.74,4.498925769073484
+1q8t,4.76,3.9675011885253912
+3acw,4.76,3.8448213159515383
+4lzs,4.8,4.171122724661253
+3r88,4.82,3.771034620791625
+4ciw,4.82,4.547881289761353
+2w4x,4.85,4.089517209735107
+2brb,4.86,4.732903124523926
+1p1q,4.89,4.193486153259274
+3d4z,4.89,3.7955817773895286
+1bzc,4.92,4.539960212866209
+1nc3,5.0,4.495808076937866
+4agq,5.01,4.714692616073604
+4w9l,5.02,4.5492668968200665
+2yge,5.06,4.502787339559934
+1e66,9.89,7.188278556848147
+1gpk,5.37,6.627061509606929
+1h23,8.35,7.640120467767334
+1mq6,11.15,8.182997659912111
+1nvq,8.25,7.991674160308843
+1o3f,7.96,7.929832432086188
+1o5b,5.77,6.377937931426995
+1oyt,7.24,7.438043994689939
+1q8u,5.96,6.246946044714352
+1r5y,6.46,6.495986522070313
+1sqa,9.21,7.132408417932131
+1u1b,7.8,6.950309609649665
+1w4o,5.22,5.631865372967537
+1yc1,6.17,7.138161297869875
+1z95,7.12,8.16023122529907
+2cet,8.02,7.4743312950378415
+2fvd,8.52,7.548192197772217
+2iwx,6.68,6.427065835626221
+2j78,6.42,6.511890819445802
+2p4y,9.0,7.3757074037475565
+2qbp,8.4,6.683465866406247
+2qbr,6.33,6.45223804780273
+2v7a,8.3,7.659672416650389
+2vvn,7.3,8.04473292189331
+2vw5,8.52,6.655207552044676
+2wca,5.6,6.835288710302733
+2weg,6.5,6.758468464373772
+2wtv,8.74,8.096293521722412
+2x00,11.33,8.133705259179687
+2xb8,7.59,6.829767467089842
+2xbv,8.43,8.429342031884767
+2xnb,6.83,7.912239552307128
+2xys,7.42,7.0454130675842315
+2y5h,5.79,7.466176282324223
+2yfe,6.63,6.686462293432614
+2yki,9.46,6.548638507910159
+2zcq,8.82,6.621591004406737
+2zcr,6.87,6.881761762066647
+3ag9,8.05,6.465643539935302
+3b68,8.4,8.190354584289548
+3coy,6.02,7.256264232788091
+3dd0,9.0,8.898755427398681
+3e93,8.85,7.927928440600588
+3ebp,5.91,5.834585341717523
+3ehy,5.85,7.041906984350589
+3ejr,8.57,7.275660043749999
+3f3c,6.02,6.169969203448485
+3f3e,7.7,6.9496301306030235
+3fv1,9.3,9.0139474576172
+3g0w,9.52,8.308721103222656
+3gbb,6.9,7.010007227856446
+3ge7,8.7,7.691785282739258
+3gnw,9.1,7.789149969110104
+3gy4,5.1,5.642898486169434
+3jvs,6.54,6.395871287023924
+3k5v,6.3,6.755995747637935
+3myg,10.7,7.480339961962891
+3n86,5.64,6.87884320557861
+3nw9,9.0,7.80833524720459
+3oe5,6.88,7.384849244354249
+3pww,7.32,7.206603011730956
+3ueu,5.24,6.669425869512939
+3uex,6.92,6.248880837237545
+3uo4,6.52,7.3048854471313485
+3uri,9.0,7.043507076727296
+3utu,10.92,9.076639426464844
+4de1,5.96,5.965974482391352
+4djv,6.72,6.763027065606689
+4gid,10.77,7.850791482153319
+4tmn,10.17,6.899314961224363
+1eby,9.7,9.071972182458495
+1g2k,7.96,9.015022155181883
+1gpn,6.48,6.737561861480713
+1h22,9.1,7.832304157910154
+1k1i,6.58,6.706198623937991
+1lpg,7.09,8.34666724058838
+1nc1,6.12,8.698183402862547
+1o0h,5.92,6.212684369488523
+1owh,7.4,6.82366798342285
+1p1n,6.8,6.3699154938171345
+1pxn,7.15,6.7172147785705585
+1qf1,7.32,7.0326707888916005
+1qkt,9.04,8.15731841669922
+1s38,5.15,6.428597907104489
+1syi,5.44,6.6256740905090314
+1y6r,10.11,8.702975983715817
+1ydr,5.52,6.023821814739987
+1ydt,7.32,6.731516535614017
+1z6e,9.72,9.223851026629642
+1z9g,5.64,6.090745917303456
+2al5,8.4,7.0378992705017165
+2br1,5.14,6.51611666404419
+2c3i,7.6,7.345220322833253
+2cbv,5.48,6.931224698944094
+2fxs,6.06,6.579123080712893
+2j7h,7.19,6.763879922875976
+2p15,10.3,8.12079060496826
+2pog,9.54,8.011308512695315
+2qbq,7.44,6.769573812530515
+2qe4,7.96,7.956994184594728
+2qnq,6.11,7.065995957580567
+2r9w,5.1,6.314880683123778
+2vkm,8.74,7.754257847521972
+2wer,7.05,7.076451121698001
+2wn9,8.52,6.491593399285888
+2wnc,6.32,6.890702031243896
+2wvt,6.12,6.788617755456541
+2xii,7.2,7.50235239619751
+2xj7,6.66,7.24886094121704
+2zb1,6.32,7.899370414654539
+2zda,8.4,7.590816088482666
+2zy1,7.4,6.625800997259521
+3arp,7.15,7.616060122827148
+3arq,6.4,7.252470330157471
+3arv,5.64,6.603162358056644
+3ary,6.0,6.32991121161499
+3b1m,8.48,6.975845003326417
+3b27,5.16,6.659595438983153
+3b5r,8.77,8.377510907250972
+3b65,9.27,7.818731971026608
+3bgz,6.26,6.53960380723267
+3bv9,5.36,7.568603907348635
+3cj4,6.51,6.248213882629393
+3coz,5.57,6.856269945117189
+3dx2,6.82,7.036528105249025
+3e5a,8.23,7.579197435076909
+3e92,8.0,7.955468993945312
+3f3d,7.16,6.852962016766357
+3fur,8.0,7.274226264727783
+3fv2,8.11,7.611759847460945
+3gc5,7.26,6.498214943859865
+3jvr,5.72,6.534867458081054
+3jya,6.89,6.210371053778073
+3kr8,8.1,7.514179323492433
+3n76,6.85,6.923076480364987
+3nx7,8.1,7.754999334240722
+3o9i,11.82,10.571575213317876
+3oe4,7.47,7.584636740136718
+3ozs,5.33,7.618785433905028
+3p5o,7.3,7.040106249493409
+3prs,7.82,7.061791739733884
+3pyy,6.86,6.777588291430662
+3qgy,7.8,7.521549591198731
+3qqs,5.82,5.700895336437992
+3rlr,7.52,7.24726395013428
+3ryj,7.8,7.819236793084718
+3syr,5.1,5.998993099548338
+3tsk,7.17,8.039678021966544
+3u5j,5.61,6.672076888208007
+3u8k,8.66,7.536642004400634
+3u8n,10.17,7.756551432366942
+3uev,5.89,6.662140793658446
+3uew,6.31,6.42144006773071
+3ui7,9.0,7.683850057873535
+3up2,7.4,7.815783694757076
+3uuo,7.96,7.269778711804197
+3wtj,6.53,7.083982210357666
+3wz8,5.82,6.306025235314942
+3zdg,7.1,6.93934756566772
+3zso,5.12,6.866331244750977
+4cra,7.22,7.864527570416258
+4crc,8.72,8.009984481304931
+4de3,5.52,6.368543999426265
+4dld,5.82,6.85392163234253
+4dli,5.62,6.715533710498047
+4e5w,7.66,8.167620159588623
+4e6q,8.36,7.694266949725342
+4ea2,6.44,7.024924207910157
+4eo8,8.15,7.719084939398188
+4eor,6.3,7.258288448657219
+4f09,6.7,7.021392902545162
+4f2w,11.3,9.941376510070796
+4f3c,11.82,10.949540416296388
+4f9w,6.94,7.133326093872067
+4gfm,7.22,6.852683067993161
+4gkm,5.17,5.5801841756713895
+4gr0,9.55,8.837895747052007
+4hge,7.92,7.7040420236450196
+4ih7,5.24,6.380635533618165
+4ivb,8.72,8.618383055297848
+4ivc,10.0,8.455736893969727
+4ivd,9.52,8.182122064904785
+4j21,7.41,6.959935733471679
+4j28,5.7,6.786026134820559
+4j3l,7.8,7.428130288195794
+4jfs,5.27,6.757419265069585
+4jia,9.22,7.775911537829589
+4k18,8.96,8.670148361224367
+4k77,6.63,7.700984968170163
+4kzq,6.1,6.754642571972658
+4kzu,6.5,6.805336975384526
+4m0y,6.46,6.64111072616577
+4m0z,5.19,7.040763745886235
+4mme,6.5,6.4431962990295455
+4ogj,6.79,6.750924728326414
+4pcs,7.85,6.55677436200562
+4qac,9.4,7.925974257635493
+4qd6,8.64,7.134632056262205
+4rfm,10.05,8.291213344226074
+4twp,10.0,8.140951649804691
+4ty7,9.52,7.714599139434814
+4w9h,6.73,6.400476244042966
+4w9i,5.96,6.039873965832514
+4wiv,6.26,6.366264964404298
+4x6p,8.3,8.644163160906981
+5c28,5.66,6.420668611987305
+5c2h,11.09,8.308975265411373
+5dwr,11.22,9.191689653570556
+5tmn,8.04,6.161969176776122

verify_with_tdc.py

@@ -0,0 +1,108 @@
+#!/usr/bin/env python3
+"""
+MillerBind v9 & v12 — TDC Evaluator Verification
+==================================================
+
+Loads pre-computed predictions and evaluates them using
+the Therapeutics Data Commons (TDC) official Evaluator.
+
+Usage:
+    pip install PyTDC numpy pandas scipy
+    python verify_with_tdc.py
+
+Author: William Miller / BindStream Technologies
+"""
+
+import os
+import sys
+import numpy as np
+import pandas as pd
+from pathlib import Path
+
+SCRIPT_DIR = Path(__file__).resolve().parent
+PREDICTIONS_DIR = SCRIPT_DIR / "predictions"
+
+
+def evaluate_model(name, csv_path):
+    """Evaluate a model's predictions using TDC Evaluator."""
+    from tdc import Evaluator
+
+    df = pd.read_csv(csv_path)
+    y_true = df["experimental_pkd"].values
+    y_pred = df["predicted_pkd"].values
+
+    print(f"\n{'=' * 60}")
+    print(f"  {name}")
+    print(f"  CASF-2016 Scoring Power (n = {len(df)})")
+    print(f"{'=' * 60}")
+
+    results = {}
+    for metric in ["pcc", "spearman", "mae", "rmse", "r2", "mse"]:
+        evaluator = Evaluator(name=metric)
+        score = evaluator(y_true, y_pred)
+        results[metric] = float(score)
+
+    print(f"\n  {'Metric':<20s} {'Value':>10s}")
+    print(f"  {'-' * 35}")
+    print(f"  {'PCC':<20s} {results['pcc']:>10.4f}")
+    print(f"  {'Spearman rho':<20s} {results['spearman']:>10.4f}")
+    print(f"  {'MAE (pKd)':<20s} {results['mae']:>10.4f}")
+    print(f"  {'RMSE':<20s} {results['rmse']:>10.4f}")
+    print(f"  {'R²':<20s} {results['r2']:>10.4f}")
+    print(f"  {'MSE':<20s} {results['mse']:>10.4f}")
+    print(f"  {'-' * 35}")
+
+    return results
+
+
+def main():
+    print("╔════════════════════════════════════════════════════════╗")
+    print("║  MillerBind — TDC Evaluator Verification              ║")
+    print("║  BindStream Technologies                              ║")
+    print("╚════════════════════════════════════════════════════════╝")
+
+    all_results = {}
+
+    v9_path = PREDICTIONS_DIR / "casf2016_v9_predictions.csv"
+    if v9_path.exists():
+        all_results["v9"] = evaluate_model("MillerBind v9 (Ensemble)", v9_path)
+
+    v12_path = PREDICTIONS_DIR / "casf2016_v12_predictions.csv"
+    if v12_path.exists():
+        all_results["v12"] = evaluate_model("MillerBind v12 (Hard Targets)", v12_path)
+
+    # Comparison table
+    print(f"\n{'=' * 60}")
+    print("  COMPARISON WITH PUBLISHED METHODS (CASF-2016)")
+    print(f"{'=' * 60}")
+    print(f"\n  {'Method':<22s} {'PCC':>8s} {'MAE':>10s} {'Year':>6s}")
+    print(f"  {'-' * 50}")
+    for name, r, mae, year in [
+        ("AutoDock Vina", 0.604, 2.05, 2010),
+        ("RF-Score v3", 0.800, 1.40, 2015),
+        ("OnionNet-2", 0.816, 1.28, 2021),
+        ("PIGNet", 0.830, 1.21, 2022),
+        ("IGN", 0.850, 1.15, 2021),
+        ("HAC-Net", 0.860, 1.10, 2023),
+    ]:
+        print(f"  {name:<22s} {r:>8.3f} {mae:>10.2f} {year:>6d}")
+
+    if "v9" in all_results:
+        r = all_results["v9"]["pcc"]
+        mae = all_results["v9"]["mae"]
+        print(f"  {'MillerBind v9 *':<22s} {r:>8.3f} {mae:>10.3f} {'2025':>6s}")
+
+    if "v12" in all_results:
+        r = all_results["v12"]["pcc"]
+        mae = all_results["v12"]["mae"]
+        print(f"  {'MillerBind v12 *':<22s} {r:>8.3f} {mae:>10.3f} {'2025':>6s}")
+
+    print(f"  {'-' * 50}")
+    print("  * Validated using TDC Evaluator (PyTDC)")
+    print(f"\n{'=' * 60}")
+    print("  VERIFICATION COMPLETE")
+    print(f"{'=' * 60}")
+
+
+if __name__ == "__main__":
+    main()