The Human Bindome project aims to create the first comprehensive de novo protein binder database targeting the entire human proteome (~20,000 proteins). To achieve this, the team will employ BindCraft, an in-house generative protein design platform built on AlphaFold2 that achieves experimental success rates between 10–100% across diverse targets. The project aims to design binders of 90 amino acids, a size in which BindCraft has proven optimal results and that ensures efficient use of computational resources. Importantly, binders of this size are cheap to produce experimentally, a property that will enhance the accessibility, scalability, and practical impact of the Human Bindome for the scientific community. BindCraft scales efficiently across multiple GPUs, making this proteome-wide task ideally suited for EuroHPC systems. The resulting Human Bindome will be released as an open-access resource for biomedical research and therapeutic development, with planned integration into UniProt through collaboration with the Swiss Institute of Bioinformatics. The team also plans to collaborate with the Mikko Taipale lab to perform large-scale experimental binding analyses, leveraging their experience in proteome-wide validation assays.
Bruno Correia, École polytechnique fédérale de Lausanne (EPFL), Switzerland