Computational Protein Design Services

Computational protein design uses AI and physics-based methods to create novel protein sequences and structures with desired properties. Modern approaches combine deep learning models (like ProteinMPNN for sequence design and RFdiffusion for backbone generation) with structure prediction tools (like AlphaFold2) to design proteins not found in nature.

ProteinMPNN achieves 52.4% native sequence recovery compared to 32.9% for Rosetta's fixed backbone design. It also runs orders of magnitude faster (~1 second per sequence vs minutes for Rosetta) and requires less expert customization. Importantly, ProteinMPNN designs have been shown to rescue previously failed Rosetta designs.

RFdiffusion has been validated for proteins ranging from 50 to over 600 amino acids. The method can design both simple monomers and complex assemblies with up to 60 subunits. For very large proteins, modular design approaches can be used.

All designs undergo rigorous computational validation including AlphaFold2 structure prediction, pLDDT scoring (>90 for high confidence), RMSD analysis to design models, and Rosetta energy scoring. Designs with atomic-level accuracy have been validated by X-ray crystallography and cryo-EM.

Deliverables include designed sequences in FASTA format ready for gene synthesis, PDB structure files, comprehensive analysis reports with rankings and metrics, validation data, and recommendations for experimental follow-up. Full IP rights to all designed sequences are transferred to the client.

Yes. RFdiffusion can be conditioned to design proteins with binding interfaces for specific targets. By providing hotspot residues from the target, the model generates scaffolds with pockets that complement the target surface. These designs have been validated to achieve nanomolar binding affinities.

Project timelines vary based on complexity. Simple redesign projects can be completed within 1-2 weeks. Complex de novo design campaigns targeting specific functions may take 4-8 weeks. We provide detailed timelines during project scoping based on your specific requirements.

Our core service focuses on computational design and validation. We can recommend experimental partners for gene synthesis, protein expression, and functional testing. Many clients combine our designs with their internal capabilities or partner labs for experimental validation.