Improving N-Glycosylation and Biopharmaceutical Production Predictions Using AutoML-Built Residual Hybrid Models

Abstract

N-glycosylation has many essential biological roles, and is important for biotherapeutics as it can affect drug efficacy, duration of effect, and toxicity. Its importance has motivated the development of mechanistic models for quantitatively predicting the distribution of N-glycans during therapeutic protein production. Here we present a residual hybrid modeling approach that integrates mechanistic modeling with machine learning to produce significantly more accurate predictions for production of monoclonal antibodies in batch, fed-batch, and perfusion cell culture. For the largest dataset, the residual hybrid models have an average 736-fold reduction in testing prediction error. Furthermore, the residual hybrid models have lower prediction errors than the mechanistic models for all of the predicted variables in the datasets. We provide the automatic machine learning software used in this work, allowing other researchers to reproduce this work and use our software for other tasks and datasets.