New Analysis Task

Configure and launch a protein structure prediction task

Select Pipeline
Choose the AI model best suited for your analysis
Best for Folding
OpenFold 3

State-of-the-art structure prediction with unprecedented accuracy

Key Features:

  • •Multimer support
  • •Template-based modeling
  • •MSA generation
Best for Ligands
Chai

Optimized for ligand binding and small molecule interactions

Key Features:

  • •Protein-ligand docking
  • •Small molecule support
  • •Binding site analysis
Best for Embeddings
ESMFold

Protein language model for rapid embedding generation

Key Features:

  • •Ultra-fast inference
  • •Language model-based
  • •No MSA required
Data Input
Provide your protein sequence or structure file

Accepts FASTA format or plain amino acid sequence

Model Benchmarks
ESMFold performance metrics
Global Distance Test (GDT)86.7
Worse78th percentileBetter
Inference Speed0.9min
Worse95th percentileBetter
Ligand RMSD1.12Å
Worse72th percentileBetter
Universal Settings
Common parameters across all models

For reproducible results

3

Number of iterative refinement cycles

Energy minimization post-processing

512

Higher depth improves accuracy but increases compute time

Estimated Time0.9min