Generative models have emerged as a powerful tool for capturing biology. In the field of protein design, generative models have recently brought previously challenging problems within reach, such as protein monomer design, motif scaffolding, and de novo interface design. The de novo design of therapeutics however, is a more challenging problem, which still remains unsolved.
Antibodies are the largest current class of therapeutics. Their biocompatibility and specificity renders them well-suited for the treatment of a broad range of conditions. However, their rational design has proved challenging, due to data limitations and their harder-to-model interfaces.
In this talk, we set out a roadmap towards solving therapeutic antibody design. We will leverage previous work to motivate this path, and survey recent developments in the field of generative modeling that we believe are relevant for this task.
Link to join virtually: https://cam-ac-uk.zoom.us/j/87421957265
This talk is not being recorded