For AI and ML practitioners, the field is shifting from passive structure prediction toward active molecular design, which raises new priorities for data, evaluation, and integration with wet-lab workflows. The technical bar moves from per-structure accuracy to end-to-end design metrics such as binding affinity prediction, manufacturability, and experimental validation throughput.
Google DeepMind’s AlphaFold transformed protein-structure prediction, and the company documents that AlphaFold 3 extends predictions to interactions among proteins, DNA, RNA, ligands, and other biomolecules (Google DeepMind, 2024), the scraped blog reports. The blog also reports that Isomorphic Labs introduced the Isomorphic Drug Design Engine (IsoDDE), which the company describes as combining protein-ligand structure prediction, binding-affinity estimation, cryptic-pocket identification, and antibody-antigen interaction modeling to support drug discovery (Isomorphic Labs, 2026).