The prediction of robust process development (PD) modalities for biologics purification is of paramount significance in the drug development process. Now that advanced computational tools have been developed, computational modeling can inform and expedite the development of biologics. The selection of appropriate purification (capture and polishing) modalities is governed by the biophysical properties of the associated variants and impurities (e.g., aggregates, host cell protein etc) to be purified. In this study, in-silico biophysical properties of a diverse set of monoclonal antibodies (~400 unique molecules) were calculated from the secondary structure (determined either experimentally or via homology modelling). For protein biophysical properties that vary as a function of protein conformation, multiple structural conformations were generated to introduce heterogeneity in the calculated properties. This allowed for ensemble averaging of the calculated properties. Further, protein surface properties (hydrophobic, positive, and negative patches) were calculated for the CDR loops, Fab, and entire antibody. The calculated in-silico properties were compared to experimental data for each antibody molecule to determine correlated parameters. Using this relational database and correlation matrix, a Quantitative Structure Property Relation (QSPR) model was developed to derive a relationship between in-silico biophysical properties and PD outcome. A directed PD approach based on a more detailed biophysical characterization of the effect of ligands on biologics biophysical properties is required to ensure successful advancement of the currently developed model. This model and principles derived from this study can be deployed to identify the platform fit and expedite the discovery of lead candidates. Further, appropriate modalities for biologics purification can be selected or prioritized based on in-silico biophysical properties of lead candidates. Principles from this study are readily applicable to chromatographic, formulation, and developability assessment of lead candidates to speed the development of molecules from DNA to first-in-human (FIH). Hence, implementation of this computational modeling tool at the onset will significantly enhance the success of biologics development from discovery, through purification to formulation activities.

Optimization of biologics process development enabled via data mining and quantitative structure analysis

255th National Meeting (2018)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.