Readiness
- Readiness
- Production Ready
- Smoke
- passed
- Artifacts
- passed
- Production
- ready
- Storage
- <5
- Strategy
- shared python312_slim.sif + isolated chembl_fpsim2_py312_site CPU Python layer + official ChEMBL 37 FPSim2 H5 database
- Next
- Production Ready for FPSim2 similarity search from SMILES with threshold/max_hits controls. Optional hardening: add exact-search mode and regenerate/pin the FPSim2 database with the same RDKit version as runtime.
No obvious large model/database dependency identified; validate upstream docs before install. 2026-07-05 SOP batch recheck: backend inventory found no staged ChemBL/FPSim2 source, HDF5 database, SIF, artifacts, or runner. Upstream ChEMBL FPSim2 database is small enough for the shared SIF/overlay path once staged (current chembl_37.h5 about 314 MB), but no direct Slurm scientific smoke has run. Kept Smoke Pending. 2026-07-05 18:19 +08:00 backend re-audit: no staged ChemBL/FPSim2 source, HDF5 database, SIF, artifacts, or runner found; upstream ChEMBL FPSim2 database remains small enough for shared SIF/overlay path, so no Need >5GB reclass. Agents agreed Smoke Pending remains correct. 2026-07-05 22:00 +08:00 backend implementation pass: official ChEMBL FPSim2 database chembl_37.h5 downloaded from EBI latest release to /media/nik/seagate_nik/bio_server/databases/chembl/chembl_37.h5 (315 MB). FPSim2/h5py/tables target install attempted but pip wheel transfers repeatedly broke before writing usable packages. Kept Smoke Pending until FPSim2 runtime imports and aspirin similarity Slurm smoke pass. 2026-07-05 23:42 +08:00 implementation pass: completed FPSim2 runtime staging at /media/nik/seagate_nik/bio_server/envs/chembl_fpsim2_py312_site with FPSim2 0.7.4, h5py 3.16.0, tables 3.11.1, RDKit 2026.03.3, and NumPy 2.5.1. Official ChEMBL 37 FPSim2 H5 database is staged at /media/nik/seagate_nik/bio_server/databases/chembl/chembl_37.h5. Direct Slurm job 434 ran aspirin similarity search at threshold 0.7 under /media/nik/seagate_nik/bio_server/runs/chembl-search/smoke_20260705_234200, returning 9 hits and writing summary.json, results.csv, job.log, slurm.out, and slurm.err. Runtime warning: the H5 was created with RDKit 2022.09.4 while installed runtime is RDKit 2026.03.3; adapter hardening should either pin matching RDKit or document/validate fingerprint compatibility. 2026-07-07T20:54:38+08:00 SOP production validation: added real runner adapter for chembl-search, deployed to /home/nik/bio_mega_runner with shared python312_slim.sif plus isolated /media/nik/seagate_nik/bio_server/envs/chembl_fpsim2_py312_site and official ChEMBL 37 FPSim2 H5. Runner API job 5e216bd732d0 / Slurm job 511 completed in 8.113 s for aspirin SMILES at threshold 0.7, returning 9 hits with top ChemBL IDs 1078517, 1280, and 1499576 at similarity 1.0. Artifacts verified: summary.json, results.csv, chembl_query.smi, chembl_runtime.json, chembl.stdout, chembl.stderr, job.log, slurm.out, slurm.err. No stderr; RDKit database-version compatibility warning is documented in runtime metadata.