Principles of Biophysical Fitness Landscape Design

We introduce foundational principles for designing customizable fitness landscapes for proteins. We focus on the problem of designing antibody sequences to create evolutionary traps to constrain the mutation of viral surface proteins and prevent the fixation of new variants. We derive an analytical relationship between antibody-antigen binding affinities and a viral variant's fitness and show that the space of all possible fitness landscapes for a chosen set of sequences separates into designable and undesignable regions. We then demonstrate that a target fitness landscape for those sequences can be obtained from stochastic optimization. Chemical reaction dynamics simulations of virions, host cells, and designed antibodies are then performed to simulate viral sequence evolution. Our novel method for fitness estimation from genotype time series data is then used to close the design loop and compare the measured fitnesses of variants with their targeted fitnesses.