AF2Dock

30-60 minutesPaperCodeNeed >5GB

AlphaFold2-derived diffusion model for protein-protein docking. Upload receptor and ligand structures; outputs docked PDBs and ipTM scores per sample.

Readiness

Readiness: Need >5GB
Smoke: failed
Artifacts: partial
Production: need_gt5gb_af2dock_openfold_cuda_weights_runtime
Storage: 5-20GB
Strategy: CUDA devel SIF with OpenFold plus AF2Dock Zenodo weights
Next: Build or pull a CUDA devel SIF with OpenFold, download AF2Dock Zenodo weights, then rerun a tiny GPU docking benchmark before wiring a runner.

Production blocker evidence 2026-07-01: source https://github.com/Graylab/AF2Dock revision dde6a4f7f14fb93ccc79101203348d41571f8f9e staged under /media/nik/seagate_nik/bio_server/tools/af2dock with OpenFold source clone attempt at be2ec1841f16c966c65ae0e7599ebbadc725757d. GPU Slurm probe job 294 completed and confirmed RTX 5050 8GB visible, but runtime prerequisites are missing: OpenFold env not installed, host lacks nvcc, AF2Dock weights not downloaded, attempted CUDA PyTorch SIF pull exceeded 18 minutes with no final SIF. No scientific docking output was produced. 2026-07-07 17:10 +0800: SOP Production Blocked audit: moved to Need >5GB because AF2Dock lacks OpenFold env, nvcc, AF2Dock weights, and a completed CUDA PyTorch SIF; the GPU probe saw hardware but produced no scientific docking output.