Predictive modeling for cell culture in commercial manufacturing of biotherapeutics.

The biopharmaceutical industry continually seeks advancements in the commercial manufacturing of therapeutic proteins, where mammalian cell culture plays a pivotal role. The current work presents a novel data-driven predictive modeling application designed to enhance the efficiency and predictability of cell culture processes in biotherapeutic production. The capability of the cloud-based digital data science application, developed using open-source tools, is demonstrated with respect to predicting bioreactor potency from at-line process parameters over a 5-day horizon. The uncertainty in model's prediction is quantified, providing valuable insights for process control and decision-making. Model validation on unseen data confirms the model's robust generalizability. An interactive dashboard, tailored to process scientist's requirements is also developed to streamline biopharmaceutical manufacturing processes, ultimately leading to enhanced productivity and product quality.

Digital application for drug product potency target evaluation in biopharmaceutical manufacturing.

Biopharmaceutical manufacturing entails a series of highly regulated steps. The manufacturing of safe and efficacious drug product (DP) requires testing of critical quality attributes (CQAs) against specification limits. DP potency concentration, which measures the dosage strength of a particular DP, is a CQA of great interest. In order to minimize the DP potency out-of-specification (OOS) risk, sterile fill finish (SFF) process adjustments may be needed. Varying the potency targets can be one such process adjustment. To facilitate such evaluation, data acquisition and statistical calculations are required. Regularly conducting the OOS risk assessment manually using commercial statistical software can be tedious, error-prone, and impractical, especially when several alternate potency targets are under consideration. In this work, the development of a novel framework for OOS risk assessment and deployment of cloud-based statistical software application to facilitate the risk assessment are presented. This application is intended to streamline the assessment of alternate potency targets for DP in biologics manufacturing. The major aspects of this potency targeting application development are presented in detail. Specifically, data sources, pipeline, application architecture, back-end and front-end development as well as application verification are discussed. Finally, several use cases are presented to highlight the application's utility in biologics manufacturing.

Application of machine learning methods to pathogen safety evaluation in biological manufacturing processes.

The production of recombinant therapeutic proteins from animal or human cell lines entails the risk of endogenous viral contamination from cell substrates and adventitious agents from raw materials and environment. One of the approaches to control such potential viral contamination is to ensure the manufacturing process can adequately clear the potential viral contaminants. Viral clearance for production of human monoclonal antibodies is achieved by dedicated unit operations, such as low pH inactivation, viral filtration, and chromatographic separation. The process development of each viral clearance step for a new antibody production requires significant effort and resources invested in wet laboratory experiments for process characterization studies. Machine learning methods have the potential to help streamline the development and optimization of viral clearance unit operations for new therapeutic antibodies. The current work focuses on evaluating the usefulness of machine learning methods for process understanding and predictive modeling for viral clearance via a case study on low pH viral inactivation.

Single-cell behavior and population heterogeneity: solving an inverse problem to compute the intrinsic physiological state functions.

The dynamics of isogenic cell populations can be described by cell population balance models that account for phenotypic heterogeneity. To utilize the predictive power of these models, however, we must know the rates of single-cell reaction and division and the bivariate partition probability density function. These three intrinsic physiological state (IPS) functions can be obtained by solving an inverse problem that requires knowledge of the phenotypic distributions for the overall cell population, the dividing cell subpopulation and the newborn cell subpopulation. We present here a robust computational procedure that can accurately estimate the IPS functions for heterogeneous cell populations. A detailed parametric analysis shows how the accuracy of the inverse solution is affected by discretization parameters, the type of non-parametric estimators used, the qualitative characteristics of phenotypic distributions and the unknown partitioning probability density function. The effect of finite sampling and measurement errors on the accuracy of the recovered IPS functions is also assessed. Finally, we apply the procedure to estimate the IPS functions of an E. coli population carrying an IPTG-inducible genetic toggle network. This study completes the development of an integrated experimental and computational framework that can become a powerful tool for quantifying single-cell behavior using measurements from heterogeneous cell populations.

A novel assay based on fluorescence microscopy and image processing for determining phenotypic distributions of rod-shaped bacteria.

Cell population balance (CPB) models can account for the phenotypic heterogeneity that characterizes isogenic cell populations. To utilize the predictive power of these models, however, we must determine the single-cell reaction and division rates as well as the partition probability density function of the cell population. These functions can be obtained through the Collins-Richmond inverse CPB modeling methodology, if we know the phenotypic distributions of (a) the overall cell population, (b) the dividing cell subpopulation, and (c) the newborn cell subpopulation. This study presents the development of a novel assay that combines fluorescence microscopy and image processing to determine these distributions. The method is generally applicable to rod-shaped cells dividing through the formation of a characteristic constriction. Morphological criteria were developed for the automatic identification of dividing cells and validated through direct comparison with manually obtained measurements. The newborn cell subpopulation was obtained from the corresponding dividing cell subpopulation by collecting information from the two compartments separated by the constriction. The method was applied to E. coli cells carrying the genetic toggle network with a green fluorescent marker. Our measurements for the overall cell population were in excellent agreement with the distributions obtained via flow cytometry. The new assay constitutes a powerful tool that can be used in conjunction with inverse CPB modeling to rigorously quantify single-cell behavior from data collected from highly heterogeneous cell populations.