A Guided Design Framework for the Optimization of Therapeutic-like Antibodies

Antibodies are a highly desirable class of protein therapeutics. However, their successful commercialization depends on meeting stringent developability criteria. To prevent developability issues early in a drug campaign, it can be advantageous to select candidates with biophysical properties characteristic of clinical stage antibodies. Inspired by this approach, we propose TherAbDesign, a general sequence-based framework that evaluates and optimizes antibodies for developability, circumventing the need for structure prediction and physics-based computation. By evaluating structure-based filtering methods on experimental datasets, using viscosity as proof-of-concept, we define a set of informative optimization objectives (charge and hydrophobicity) for guided design. TherAbDesign proposes rational modifications to mimick the biophysical properties of successful therapeutic antibodies. We show that this method improves known developability liabilities, such as viscosity shown here, without explicitly modeling their mechanism of action through design guidance on biophysical properties.