General protein complex
One or more protein chains modeled together in a single complex.
Start this methodPredict and inspect a biomolecular complex
Predict an all-atom complex, then inspect the structure, confidence, entity conservation, and geometry checks together.
Choose an approach
Scientific approaches
Select the analysis that best matches your scientific question. Each approach opens with its relevant inputs and controls.
Antibody–antigen complex
Heavy or VHH, optional light chain, and antigen using the separate antibody–antigen checkpoint.
Start this methodStructure prediction
How might these molecules assemble in 3D?
Describe one protein complex or antibody–antigen system, then validate the molecular inputs before starting the prediction.
Expected output A predicted 3D complex and result summary saved with the selected Project.
Molecule 1
Protein-only complexes are the best-tested mode. Review mixed nucleic-acid, ligand, and ion predictions especially carefully.
Advanced prediction settings
Fixed prediction protocol
1 sample · 200 diffusion steps · 10 pairformer cycles · fp32 · deterministic
Analysis workspace
No setup is required. Ubi saves the predicted structure and result summary to an editable Project, creating a default one after sign-in only if needed.
No Project setup is needed now. Sign in when you preview, and Ubi will use your current analysis Project or create a default one.
Prepare
Molecular entities for one complex
Define one complex with up to eight protein, DNA, RNA, ligand, or ion entities. Protein-only and antibody–antigen inputs are available now.
- Exactly one complex and at most eight entities
- At most 512 total polymer residues/bases including copies
- Explicit one-based entity, copy, residue, modification, and covalent-bond references
General protein entity
{ proteinChain: { sequence, count: 1 } }Antibody–antigen
heavy/VHH + optional light + antigenHow the analysis starts
Enter the scientific inputs above. Sign-in is requested only when you preview or submit, and Ubi saves the analysis to your current or default Project.
Configure the scientific method
Review the available method settings here and use the input panel above to validate them against your sequences.
Checkpoint
General or antibody–antigen
Choose the model that matches the molecular system.
Inference profile
1 sample · 200 steps · 10 recycles
Use the standard deterministic settings for comparable predictions.
Seed
Explicit deterministic seed
Repeat the same configured prediction or vary the seed deliberately.
Review results as scientific outputs
Results open with the figures, structures, sequences, and metrics needed to answer the scientific question. Downloadable files remain available for downstream analysis.
How to interpret the result
- pLDDT summarizes local model confidence; pTM summarizes global fold confidence; interface pTM is useful only for multichain interfaces.
- Confidence scores are model evidence, not a probability that an interaction occurs or an affinity estimate.
- Inspect low-confidence loops, interfaces, clashes, and geometry alongside experimental or orthogonal structural evidence.
1
Interactive all-atom structure
Inspect the predicted mmCIF structure directly in the shared Mol* viewer.
2
Confidence summary
Review pLDDT, pTM, interface pTM, predicted distance error, clashes, and chain metrics.
3
Structure validation
Confirm entity sequence conservation, finite coordinates, atom/residue counts, and backbone geometry.
Continue your analysis
Use a complementary method on the same Project data.
Method scope and limitations
- The upstream OpenDDE project is labeled preview; interpret predictions accordingly.
- Templates, external MSA search, alternate inference settings, and multiple complexes are not supported in this analysis.
- Each analysis predicts one complex, and larger complexes may take many minutes.
