Protein Design for Pandemic Preparedness

Improving the speed of a pandemic response provides an exceptional return on investment, in terms of human life, health, and economic well-being.

While, with modern vaccine platforms, the "design" component of vaccine discovery can claim to be almost instantaneous, with the sequence of many SARS-CoV-2 vaccines being decided in mere days, the same is not true of therapeutics, where discovery can take months.

This proposal aims to address this delay.Recent developments in generative AI, which are powering everything from writing assistants to AI art, are also disrupting computational protein design.

In the event of an emerging pandemic, we propose to use generative protein design approaches, both ours and others, to almost-instantaneously "repurpose" antibodies from a related (and well-described) "prototype" pathogen, redesigning key regions of the antibody sequence to maximize affinity against the emerging virus.

We argue that this is within reach of the technology.The project spans computational methods development to experimental validation.

We will develop various computational protein design approaches tailored to this task, and explore approaches from other groups, and characterize the extent to which repurposing antibodies succeeds, with divergence between prototype and emerging antigen.

Finally, we will build an end-to-end pipeline which can, after an emerging virus is first sequenced, mine the PDB and generate a number of candidate repurposed antibodies in minutes.