AlphaProtein Novo: generative diffusion pipeline for de novo enzyme design (github.com)

AlphaProtein Novo (AP Novo) has introduced a generative diffusion pipeline designed for de novo enzyme design through structural motif scaffolding. This innovative system employs a diffusion model that simultaneously generates protein structures and sequences, tailored to specific catalytic motifs and ligand contexts. The comprehensive pipeline integrates features such as sequence redesign using LigandMPNN, structure prediction via AlphaFold 3, and evaluation metrics, all orchestrated through a single script. The release encourages researchers to cite their foundational paper when disclosing findings related to the use of this code. The significance of AP Novo lies in its ability to streamline the enzyme design process, facilitating the creation of proteins that can perform novel reactions with non-natural substrates. The pipeline's modular architecture supports efficient research workflows, allowing customization of input configurations and design objectives. Key technical components include the use of JAX for diffusion generation and evaluation, along with options for PyTorch-based sequence design. Researchers can leverage this tool to explore diverse enzyme functionalities, potentially transforming applications in synthetic biology and biotechnology.