Readiness
- Readiness
- Production Ready
- Smoke
- passed
- Artifacts
- passed
- Production
- production_pass
- Storage
- <5
- Strategy
- shared python312_slim.sif + sequence_design_torch_cpu_py312 ProteinMPNN/PyTorch layer + bundled CyclicMPNN weights
- Next
- Production Ready. Optional hardening: expose fixed-position/omit-AA/tied-position JSON controls and larger batch presets.
No obvious large model/database dependency identified; validate upstream docs before install. 2026-07-05 SOP batch recheck: backend inventory found no staged CyclicMPNN source/weights/artifacts/runner. It matches the already validated ProteinMPNN shared-SIF pattern and includes cyclicmpnn_weights/cyclicmpnn_48_010.pt upstream. Kept Smoke Pending. 2026-07-05 18:24 +08:00 backend re-audit: no staged CyclicMPNN source, weights, artifacts, or runner found; only local form/spec evidence. Kept Smoke Pending. 2026-07-06 implementation pass: cloned ParisaH-Lab/CyclicMPNN at /media/nik/seagate_nik/bio_server/tools/cyclicmpnn/source/CyclicMPNN rev a6931519296247a9062eb0508dbc70430ce9f0de. Repo includes bundled cyclicmpnn_weights/cyclicmpnn_48_010.pt and example PDBs. Direct Slurm job 456 passed at /media/nik/seagate_nik/bio_server/runs/cyclicmpnn/smoke_20260706_160000 using shared python312_slim.sif plus sequence_design_torch_cpu_py312 and the bundled 5L33 PDB. The smoke forced --path_to_model_weights cyclicmpnn_weights and --model_name cyclicmpnn_48_010, generated 2 designed sequences plus native FASTA record, finite scores 0.7743-1.9672, and artifacts summary.json, results.csv, designed_sequences.csv, cyclicmpnn_out/seqs/5L33.fa, cyclicmpnn_out/scores/5L33.npz, stdout/stderr, job.log, slurm.out, and empty slurm.err. 2026-07-08T02:06:39+08:00: Production Ready promotion: backend runner API job 640d7c90e901 / Slurm job 534 completed cleanly using shared python312_slim.sif, sequence_design_torch_cpu_py312, bundled CyclicMPNN source/weights, and bundled 5L33 fallback PDB. Produced zero-byte slurm.out/slurm.err, summary.json, designed_sequences.csv/fasta, design_scores.npz, and finite design scores for two generated sequences.