AI proved that scientists were looking at the wrong part of antibodies to figure out how stable they are.
April 14, 2026
Original Paper
Biologically-Grounded Multi-Encoder Architectures as Developability Oracles for Antibody Design
arXiv · 2604.09369
The Takeaway
While researchers thought stability was a complex interaction, the AI found that a single chain is almost entirely responsible for antibodies clumping together. This 'oracle' is now overturning long-standing human assumptions about how to design better drugs.
From the abstract
Generative models can now propose thousands of \emph{de novo} antibody sequences, yet translating these designs into viable therapeutics remains constrained by the cost of biophysical characterization. Here we present CrossAbSense, a framework of property-specific neural oracles that combine frozen protein language model encoders with configurable attention decoders, identified through a systematic hyperparameter campaign totaling over 200 runs per property. On the GDPa1 benchmark of 242 therape