Elution relationships to model affinity chromatography using a general rate model

Physics-informed neural networks to solve lumped kinetic model for chromatography process

Numerical method is widely used for solving the mechanistic models of chromatography process, but it is time-consuming and hard to response in real-time. Physics-informed neural network (PINN) as an emerging technology combines the structure of neural network with physics laws, and is getting noticed for solving physics problems with a balanced accuracy and calculation speed. In this research, a proof-of-concept study was carried out to apply PINN to chromatography process simulation. The PINN model structure was designed for the lumped kinetic model (LKM) with all LKM parameters. The PINN structure, training data and model complexity were optimized, and an optimal mode was obtained by adopting an in-series structure with a nonuniform training data set focusing on the breakthrough transition region. A PINN for LKM (LKM-PINN) consisting of four neural networks, 12 layers and 606 neurons was then used for the simulation of breakthrough curves of chromatography processes. The LKM parameters were estimated with two breakthrough curves and used to infer the breakthrough curves at different residence times, loading concentrations and column sizes. The results were comparable to that obtained with numerical methods. With the same raw data and constraints, the average fitting error for LKM-PINN model was 0.075, which was 0.081 for numerical method. With the same initial guess, the LKM-PINN model took 160 s to complete the fitting, while the numerical method took 7 to 72 min, depending on the fitting settings. The fitting speed of LKM-PINN model was further improved to 30 s with random initial guess. Thus, the LKM-PINN model developed in this study is capable to be applied to real-time simulation for digital twin.

Predictive scaling of fiber-based protein A capture chromatography using mechanistic modeling

Protein A affinity chromatography is an important step in the purification of monoclonal antibodies (mAbs) and mAb-derived biotherapeutics. While the biopharma industry has extensive expertise in the operation of protein A chromatography, the mechanistic understanding of the adsorption/desorption processes is still limited, and scaling up and scaling down can be challenging because of complex mass transfer effects in bead-based resins. In convective media, such as fiber-based technologies, complex mass transfer effects such as film and pore diffusions do not occur which facilitates the study of the adsorption phenomena in more detail and simplifies the process scale-up. In the present study, the experimentation with small-scale fiber-based protein A affinity adsorber units using different flow rates forms the basis for modeling of mAb adsorption and elution behavior. The modeling approach combines aspects of both stoichiometric and colloidal adsorption models, and an empirical part for the pH. With this type of model, it was possible to describe the experimental chromatograms on a small scale very well. An in silico scale-up could be carried out solely with the help of system and device characterization without feedstock. The adsorption model could be transferred without adaption. Although only a limited number of runs were used for modeling, the predictions of up to 37 times larger units were accurate.

Chromatography bioseparation technologies and in-silico modelings for continuous production of biotherapeutics

The potential of continuous bioprocessing is hindered by the bottlenecks of chromatography processing, which continues to be executed in batch mode. Highlighting the critical drawbacks of batch chromatography, this review underscores the transition that the industry has made by implementing continuous upstream process without devising a working model for downstream chromatography operations. Even though multitude of process development initiatives have commenced, the review emphasizes the first principle models of chromatography on which these initiatives are built. Various models of continuous chromatography, which are essential, but not limited to multi-column systems, employed to congeal a unified process are reviewed. Advancements made by several mechanistic models and simulations to maximize productivity and performance are described, in an attempt to provide the integral tools. The modeling tools can be used for development of a strong model based control strategy and can be embedded into the continuous chromatography framework. The review addresses the limitations and challenges of the current modeling methods for development of robust mechanistic modeling and efficient unit operation platform in continuous chromatography.