CombFold

NewPaperCodeNeed >20GB

End-to-end structure prediction of large protein complexes: AlphaFold-Multimer on subunit pairs plus combinatorial assembly, from sequence

Readiness

Readiness: Need >20GB
Smoke: not_tested
Artifacts: not_applicable
Production: need_gt20gb_afm_colabfold_combfold_pipeline_pending
Storage: >20
Strategy: large AFM/ColabFold + CombFold pipeline storage/runtime
Next: Decide scope. For full sequence-to-complex, stage AFM/ColabFold databases/runtime and CombFold, then run bounded Slurm workflow. For assembler-only, change UI/runner contract to accept precomputed AFM PDBs plus subunits.json and smoke repo example.

No obvious large model/database dependency identified; validate upstream docs before install. 2026-07-05 SOP batch recheck: backend inventory found no staged CombFold source/AFM predictions/runtime/artifacts/runner. Current sequence-input UI implies sequence-to-complex workflow, while upstream CombFold requires AlphaFold-Multimer/ColabFold subcomplex predictions; assembler-only smoke would not validate advertised tool. Reclassified from Smoke Pending to Need >20GB.