ProteinMPNN-ddG Binder

1-5 minutesPaperCodeProduction Ready

Predict stability of point mutations

Readiness

Readiness: Production Ready
Smoke: passed
Artifacts: passed
Production: production_pass
Storage: <5
Strategy: shared PyTorch SIF + tool ext3 overlay; avoid per-tool torch duplication
Next: Production Ready. Optional next hardening: expose binder-specific chain/interface controls in the UI.

No obvious large model/database dependency identified; validate upstream docs before install. Production evidence 2026-07-01: source https://github.com/PeptoneLtd/proteinmpnn_ddg revision 98302b5 staged for binder at /media/nik/seagate_nik/bio_server/tools/proteinmpnn-ddg-binder/source/proteinmpnn_ddg. Worker Slurm job 277 passed using upstream CPU SIF /media/nik/seagate_nik/bio_server/containers/tools/proteinmpnn_ddg_1.0.0_base_cpu.sif; produced 4300 mutation rows across 215 positions with logit_difference_ddg range -8.397502 to 6.332352. Web runner adapter was wired to runner_tool proteinmpnn-ddg-binder and submitted through cluster runner API as Slurm job 278, remote id d70ac69194c9. Runtime_seconds 5.543. Artifacts: input.pdb, proteinmpnn_ddg_predictions.csv, results.csv preview, summary.json, stdout/stderr, Slurm logs. Sanity: 4300 mutation rows, expected columns pre/post/pos/logit_difference/logit_difference_ddg; first logit_difference_ddg -2.3459811.