Readiness
- Readiness
- Production Ready
- Smoke
- passed
- Artifacts
- passed
- Production
- production_pass_nbforge_cpu_no_minimize
- Storage
- <5
- Strategy
- shared python312_slim.sif + nbforge_py312_site/palm_py312_site with host HMMER bind and staged nbforge.ckpt
- Next
- Production Ready. Optional hardening: expose minimization toggle/runtime warnings and multi-sequence batch mode in the UI.
No obvious large model/database dependency identified; validate upstream docs before install. 2026-07-05 SOP batch recheck: backend inventory found no staged NbForge source/prepare cache/runtime/artifacts/runner. Upstream needs PyTorch, OpenMM, ANARCI/mmseqs2 and nbforge prepare; plain python312_slim is insufficient but no >5GB storage proof was recorded. Kept Smoke Pending with shared runtime action. 2026-07-05 18:35 +08:00 backend re-audit: no staged NbForge source/prepare cache/runtime/artifacts/runner found; requires PyTorch/OpenMM/ANARCI/mmseqs2 plus nbforge prepare. Kept Smoke Pending. 2026-07-06 21:01 +08:00 implementation attempt: cloned sormanni-lab/nbforge to /media/nik/seagate_nik/bio_server/tools/nbforge/source/nbforge rev f5c90aa. Upstream environment.yml targets Python 3.12 with PyTorch, mmseqs2, OpenMM, PDBFixer and related dependencies; README requires nbforge prepare and ANARCI trimming before inference. No prepare cache/runtime smoke exists yet. 2026-07-06 21:34 +08:00 recheck: staged source remains present, but no ANARCI/mmseqs2/OpenMM/PDBFixer runtime or nbforge prepare cache exists; no TEST1 smoke submitted. 2026-07-06 22:18 +08:00 continuation: installed NbForge 0.1.1 no-deps into /media/nik/seagate_nik/bio_server/envs/nbforge_py312_site, added dm-tree/ml-collections/loguru/typer/absl/annotated-doc leaves, and verified bundled checkpoint /media/nik/seagate_nik/bio_server/envs/nbforge_py312_site/NbForge/weights/nbforge.ckpt (98,674,455 bytes). Direct Slurm job 474 created nbforge_infer.pt cache (33,244,762 bytes) but TEST1 prediction stopped because ANARCI is required. Host has mmseqs and hmmscan, but python312_slim lacks git/hmmscan; local ANARCI source clone rev 79f6c57 could not be wheel-installed because its build tries to write /usr/local/bin/ANARCI in the read-only SIF. No production smoke passed. 2026-07-06 23:06 +08:00 continuation: attempted manual ANARCI staging for NbForge by copying package/bin and building HMMs. Build script needed a python shim and then failed because IMGT returned HTTP 502 while downloading germline data; HMMs/germlines.py were not produced. Existing NbForge cache from Slurm job 474 remains valid, but prediction still blocked by ANARCI/HMMER runtime. 2026-07-07 00:14 +08:00 continuation: included in next-10 Smoke Pending review; no SOP-valid direct Slurm smoke completed in this pass. Status unchanged pending the previously identified runtime/model/container blocker. 2026-07-07 00:58 +08:00 continuation: retried ANARCI build with a temporary `python` shim. Pipeline got past the missing command but IMGT again returned HTTP 502 while fetching germline data; Homo_sapiens_HJ.fasta and all_js_aligned.fasta remained missing, so ALL.hmm/germlines.py were not produced. Existing NbForge checkpoint/cache remain staged, but no TEST1 smoke passed. 2026-07-07 01:16 +08:00 continuation: reviewed in remaining Smoke Pending batch; no SOP-valid direct Slurm smoke completed and status remains unchanged pending the established runtime/model/container blocker. 2026-07-07 03:24 +08:00 promotion: completed ANARCI/HMMER asset staging by copying trusted ALL.hmm and germlines.py from deeprank_ab_py312_site into nbforge_py312_site and pressing ALL.hmm with hmmpress. Direct Slurm job 490 passed at /media/nik/seagate_nik/bio_server/runs/nbforge/smoke_20260707_032323 using shared python312_slim-style host runtime layers nbforge_py312_site plus palm_py312_site, CPU mode, --no-minimize, and checkpoint /media/nik/seagate_nik/bio_server/envs/nbforge_py312_site/NbForge/weights/nbforge.ckpt (98,674,455 bytes, sha256 1762018a09b1e6f12ef68de66db67df5ceca269f3fa53bcbf0f27998d0ad3195). ANARCI trimmed TEST1 to 125 positions, model loaded 388 weights with missing=0/unexpected=0, and output predictions/TEST1.pdb plus TEST1_summary.csv. summary.json and results.csv recorded plddt_mean 89.05248666381836, plddt_hcdr3_mean 72.34318699556239, cdr3_length 17. Runner/API adapter remains pending. 2026-07-08T20:42:25+08:00: Production Ready promotion: backend runner API job 1242f4a2c36b / Slurm job 562 completed cleanly using shared python312_slim.sif, nbforge_py312_site/palm_py312_site, host HMMER bind for ANARCI, and staged NbForge checkpoint. Produced zero-byte slurm.out/slurm.err, prediction.pdb, results.csv, pLDDT mean 89.05249356079102, HCDR3 pLDDT mean 72.34318789313821, and CDR3 length 17.