Structure method
Predict and inspect a biomolecular complex
Predict an all-atom complex, then inspect the structure, confidence, entity conservation, and geometry checks together.
Method details
Input checks
- One to four ordered complexes and at most eight entities per complex
- At most 512 total polymer residues/bases including copies per complex
- Explicit one-based entity, copy, residue, modification, and covalent-bond references
Expected results
- Interactive all-atom structure — Inspect the predicted mmCIF structure directly in the shared Mol* viewer.
- Confidence summary — Review pLDDT, pTM, interface pTM, predicted distance error, clashes, and chain metrics.
- Structure validation — Confirm entity sequence conservation, finite coordinates, atom/residue counts, and backbone geometry.
Example results
Structure prediction
How might these molecules assemble in 3D?
Build one complex here, or import 1–4 complexes for sequential execution after one model load. OpenDDE predicts candidate all-atom assemblies and confidence estimates; it does not establish that the molecules interact in nature.
Expected output One predicted 3D complex per submitted job, with result summaries saved to the Project.
Protein-only complexes are the best-tested mode. Review mixed nucleic-acid, ligand, and ion predictions especially carefully.
Prediction settings
Compute budget: 2,000 / 2,000 units. Reduce samples, steps, or cycles if the limit is exceeded.
