Accelerating chromatography process development: A hybrid approach to model hydrophobic interaction column based on...

In an industrial setting, the development of hydrophobic interaction chromatography (HIC) processes for protein purification is often time-consuming and resource-intensive, requiring extensive experimentation to optimize conditions for specific protein-resin systems. This study presents a hybrid modeling framework for hydrophobic interaction chromatography, which can be used in commercial chromatography process development to address these challenges. The hybrid model integrates mechanistic principles, including fluid dynamics and mass transfer from the lumped rate model, with data-driven binding isotherms derived using high-throughput experimental platforms. Fully automated systems such as slurry plates and micropipette tips are used to characterize binding isotherms across a two-dimensional parameter space of pH and salt concentration for a monoclonal antibody. The model's predictive accuracy is validated by comparing simulated chromatograms with experimental data for target monoclonal antibodies (mAbs) and high molecular weight aggregates (HMWs). Mean squared errors of 0.45, 3.66, and 0.03 mg/mL for all proteins, target monomer, and HMW species are obtained with the proposed modelling methodology.