As one of fastest growing classes of therapeutics, monoclonal antibodies (mAbs) can exhibit low solubility and high viscosity especially when it comes to high concentration formulations, posing a challenge for mAb development, manufacturing and administration. In the early development stage, thorough experimental evaluations are material and labor intensive, while molecular dynamics simulations to assess these properties are computationally expensive. To address this dilemma, we propose a workflow for the development of predictive models to facilitate the <i>in silico</i> evaluation of lead molecules during candidate selection. A diverse internal set of mAbs with varying viscosity and solubility data formed the basis for model development. To supplement the sparse internal data set, we conducted a comprehensive evaluation of readily available data published in the literature. Importantly, we introduce a methodology to identify the most advantageous external data sets for inclusion in the model. A wide range of descriptors that serve as the surface property ‘fingerprint’ were calculated for each mAb in both the internal and external data sets. Using these surface properties as input, classification models were developed to help distinguish mAbs that showed desirable viscosity and solubility from those with poor behavior. Significantly, this work extends cluster descriptors that have previously been shown to be important in multimodal chromatography models to protein-protein interactions. Furthermore, local descriptors provided key insights into the surface charge and hydrophobicity that contributed to different viscosity and solubility behavior for several variant/wild type sets. This work sets the stage for the development of more quantitative QSAR models for the prediction of viscosity and solubility and has important implications for the design of mAbs for biomanufacturability.

<i>In silico</i> evaluation of antibody developability: QSAR models to predict mAb solubility and viscosity

2019 Spring National Meeting

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

In silico evaluation of antibody developability: QSAR models to predict mAb solubility and viscosity