Antibody developability prediction improves with shared structure and language models

Multi-Assay Antibody Developability Prediction with Shared Structure-Language Representations

Machine Learning

Summary

Predicting which antibodies will work well as medicines early on is hard because there isn’t much experimental data. The authors combined two types of information: detailed 3D structural features of antibodies and language-based models that analyze their sequences. Their combined model learns from multiple related tests simultaneously, improving accuracy for several important antibody properties. This helps scientists better guess which antibodies can become drugs before costly lab work.

What this means in practice

  • For biopharmaceutical discovery teams: Predict multiple antibody developability properties from sequence and structure to prioritize candidates early in drug development.
  • For protein engineering teams: Use combined structure-language models to improve selection of antibody variants with desirable binding and stability traits across assays.

Authors

Daria B. Kokh

Abstract

Early assessment of antibody developability is critical in drug discovery, but computational approaches remain limited by the scarcity of experimental data. We explore two strategies for working within that limit. First, we designed per-residue physicochemical descriptors encoding the local structural environment, including new terms such as directional charge asymmetry and screened electrostatic potential. These feed a convolutional network, combined with protein language model (PLM) embeddings in a two-stream architecture: the two are encoded and predicted independently and merged only at the prediction stage. Second, a shared trunk with per-assay heads passes information between related assays. Both are implemented in the Multi-Assay Developability Model (MADM) and evaluated on eight assays: HIC, heparin affinity (HAC), AC-SINS, and polyspecificity ELISA against BVP, dsDNA, insulin, ovalbumin and soluble membrane proteins (SMP), with labels from 29 open sources. Structural descriptors match or exceed PLM embeddings (AbLang2, ESM2) on several assays; combining the streams gives a small but consistent gain in accuracy and reduces seed-to-seed variance. Multi-assay training raises Spearman rho for ovalbumin and SMP by about 0.1, where assays correlate and share sequence space. Under cluster-split cross-validation the strongest results are for HAC, ovalbumin, insulin, SMP and HIC (rho = 0.70-0.82); AC-SINS, BVP and dsDNA are weaker (0.50-0.55). In these three, sequence clusters largely coincide with data source, so a held-out fold measures transfer to differently calibrated measurements as well as to new sequences. On the 80-antibody GDPa3 hold-out, performance falls by rho around 0.1 for all but SMP (0.22). Transfer to the VHH panel of Ginkgo GDPa5 gives comparable results, and scFv validation against binary SMP labels yields F1 = 0.89, AUC = 0.95.