Computational Developability Assessment of Antibody Therapeutics

Abstract

Therapeutic monoclonal antibodies and their associated biologic derivatives are a key component of the commercial clinical pipelines of the global pharmaceutical industry. The availability of large datasets of antibody biophysical properties enables the search for predictive models and computational tools for the developability assessment of drug candidates. This thesis work has evaluated the scope of using biopharmaceutical informatics approaches for the prediction of developability issues such as stability, aggregation, and immunogenicity. We firstly establish developability guidelines based on in-silico metrics used for the assessment of antibody properties and derived from clinical-stage antibodies. These new computational developability guidelines serve as benchmarks for acceptable biophysical properties desired in antibody therapeutics. We have also highlighted the developability potential of natural human immune repertoire. Our developability criteria were then utilized to compare the developability profile of major antibody discovery platforms to guide the selection of platform technologies for next-generation biotherapeutics. Next, we have used machine learning algorithms to estimate clinical trial progression of antibody therapeutics. Finally, we have validated our developability criteria performance in flagging antibodies that have caused serious adverse events or failure in clinical trials with an overall model accuracy of 80.6%. Finally, a summary and conclusion of the work are provided with a future outlook towards biopharmaceutical informatics for antibody drug discovery and optimization.

Date of Award	31 Dec 2023
Original language	English
Awarding Institution	The University of Manchester
Supervisor	James Warwicker (Supervisor) & Robin Curtis (Supervisor)