Reach Our Protein Experts:

appliedproteinservices@lonza.com

The Sentinel APART™ platform is based on a computational model predicting aggregation risk for antibodies. The model was built and validated on experimental data for over 500 full-length IgG antibodies using physicochemical characteristics of amino acids and statistical and machine-learning approaches. The development and validation of the model were previously described (Obrezanova et al. 2015).

The model provides qualitative prediction, high or low, of the aggregation risk for antibodies using the variable domain primary sequence of antibodies as input. Classification into high or low risk of aggregation corresponds to a threshold of ~5% of total soluble aggregate (as measured by SE-HPLC) In addition to predicting aggregation propensity, the model also generates an aggregation score which provides an estimate of confidence in the prediction.