TLimmuno

PaperCodeProduction Ready

Class II Immunogenicity Prediction

Readiness

Readiness: Production Ready
Smoke: passed
Artifacts: passed
Production: production_ready_cpu_tensorflow
Storage: <5
Strategy: shared TensorFlow SIF + tool ext3 overlay
Next: Validate file-mode batch input from web UI and consider an adapter option to skip rank for fast score-only jobs if users need low-latency predictions.

SOP validation used upstream TLimmuno2 revision 8b02617025e6896f2a4bcd1f977ed5b4a720805b at /media/nik/seagate_nik/bio_server/tools/tlimmuno/source. Shared Python 3.9 TensorFlow env at /home/nik/micromamba-root/envs/canya-py39. Standalone Slurm job 358 and API Slurm job 359 completed on CPU using line mode epitope GLLFRRLTSREVLLL and HLA DRB1_0803. Prediction score 0.9916319 and rank 0.00103332. Artifacts include prediction.json, results.csv, summary.json, stdout/stderr, and Slurm logs. Repo is data-heavy (~1.3G) but includes required SavedModel weights and small runtime data.