Info
Talk Summary:
Generative protein design can now produce binders against a chosen target in days, but almost none of those designs have been tested as radiopharmaceuticals in a living animal. Design scores and in vitro affinity say very little about what a molecule does once it is injected.

This talk covers what we found when we took de novo designed miniproteins all the way through radiolabeling and PET imaging in tumor-bearing mice. Using RFdiffusion, we designed small binders against HER2, screened them by ELISA and biolayer interferometry, radiolabeled the lead with copper-64, and imaged it in xenograft models, moving from computational model to in vivo image in weeks rather than years. We then repeated the exercise against a second, structurally unrelated target to ask whether the approach generalizes.

I will focus on where the computational predictions held, where they did not, and what that means for anyone deciding whether to build a discovery program on generative design.

Author
Ben Larimer
University of Alabama at Birmingham