ProteinReDiff: Complex-based ligand-binding proteins redesign by equivariant diffusion-based generative models.

Area of Science:

• Computational biology
• Protein engineering
• Drug discovery

Background:

• Proteins are essential for biological processes, interacting with ligands to perform vital functions.
• Designing effective ligand-binding proteins is crucial for advancing drug development and therapeutic strategies.

Purpose of the Study:

• To introduce ProteinReDiff, a diffusion framework for redesigning ligand-binding proteins.
• To enable the creation of high-affinity ligand-binding proteins computationally.

Main Methods:

• Utilized equivariant diffusion-based generative models.
• Leveraged initial protein sequences and ligand SMILES strings.
• Did not require detailed structural information for protein design.

Main Results:

• Demonstrated potential for creating high-affinity ligand-binding proteins.
• Evaluated performance based on sequence diversity, structural preservation, and binding affinity.
• Showcased the framework's capabilities in computational drug discovery and protein engineering.

Conclusions:

• ProteinReDiff offers a novel approach to designing functional ligand-binding proteins.
• The framework has significant implications for accelerating computational drug discovery.
• ProteinReDiff advances the field of protein engineering through innovative generative models.