Net Haskins | Scientist II, Protein Engineering
Amgen

Festival of Biologics Day 2 @ 17:30

Toward enhancement of antibody thermostability and affinity by computational design in the absence of antigen

The complexity of therapeutic antibody discovery and optimization necessitates advanced in silico tools to improve efficiency and accuracy. This study addresses the challenges associated with unreliable antibody-antigen structural predictions and the limitations of traditional docking methods. We leverage DeepAb, a deep learning model that predicts antibody Fv structures directly from sequence data, in combination with deep mutational scanning (DMS) enrichment data, to design and evaluate 200 optimized variants of an anti-hen egg lysozyme (HEL) antibody. Comprehensive analysis revealed that 91% of the variants exhibited increased thermal stability, while 94% showed improved binding affinity. Notably, 10% of the variants demonstrated 5- to 21-fold affinity enhancements and more than 2.5°C increases in thermostability, all while maintaining favorable developability profiles. Moreover, in silico analyses suggest that this approach can enrich for binding affinity without prior DMS data, circumventing the need for accurate antibody-antigen interface prediction. These findings highlight the transformative potential of deep learning models in accelerating antibody optimization, overcoming structural prediction limitations, and enabling more efficient biologics development pipelines

Net Haskins, Scientist II, Protein Engineering, Amgen

×

Net Haskins / Toward enhancement of antibody thermostability and affinity by computational design in the absence of antigen

Net Haskins, Scientist II, Protein Engineering,
Amgen

Net Haskins, M.S. is a Scientist II at Amgen, specializing in antibody discovery and optimization. She holds master’s degrees in Biological Sciences from the University of Maryland and Biochemistry and Molecular Biology from Georgetown University. Her current role involves co-leading biologics...

CloseMore