On February 10, 2026, Isomorphic Labs unveiled the Isomorphic Labs Drug Design Engine (IsoDDE), a unified computational drug design system that marks a significant leap beyond AlphaFold 3 in both predictive accuracy and practical drug discovery capabilities. The announcement represents a pivotal transition for the Alphabet-backed company, moving from structure prediction into end-to-end drug design.
Doubling AlphaFold 3's Accuracy
On the challenging 'Runs N' Poses' benchmark, designed to test out-of-distribution performance on protein-ligand structure prediction, IsoDDE more than doubles AlphaFold 3's accuracy on the hardest cases. The system successfully captures complex phenomena such as induced fit and cryptic pocket opening, even in molecular systems highly dissimilar to those seen during training. For binding affinity predictions, IsoDDE exceeds the accuracy of gold-standard physics-based methods like free energy perturbation (FEP) calculations, while operating at a fraction of the time and cost.
Breakthrough in Antibody Design
IsoDDE also demonstrated remarkable capabilities in antibody design. On a set of 334 low-similarity antibody-antigen complexes, IsoDDE generated interface structures more similar to experimentally observed structures than both AlphaFold 3 and Boltz-2, outperforming AlphaFold 3 by 2.3x and Boltz-2 by 19.8x. This represents a substantial advance in the ability to computationally predict and design antibody-antigen interactions.
Novel Binding Site Discovery
Perhaps most strikingly, IsoDDE demonstrated the ability to identify novel binding pockets on target proteins using only the amino acid sequence as input, without any prior experimental data about ligand binding. In one validation, the system recapitulated the recent discovery of a novel cryptic site on cereblon, predicting the location of both the known and the newly discovered cryptic sites using just the protein sequence, without specifying the identity of the ligands.
Path to Clinical Trials
The IsoDDE announcement comes as Isomorphic Labs prepares for its first human clinical trials. The company is staffing up for IND-enabling studies across several lead candidates in oncology and immune-mediated disorders. Experts predict the first AI-designed molecules from the company's $3 billion partnerships with Eli Lilly and Novartis could enter Phase I trials by late 2026, providing the first real-world test of whether AlphaFold-derived drug design can produce candidates that outperform those discovered through traditional means.
Industry Significance
IsoDDE represents a shift in the AI drug discovery paradigm, moving from structure prediction as a standalone tool into integrated drug design where molecular binding, selectivity, and druggability are addressed simultaneously within a single computational framework. The system's ability to predict cryptic binding sites and handle complex induced-fit dynamics suggests it could unlock targets that have historically been considered undruggable, potentially opening new therapeutic frontiers in oncology, neurodegeneration, and infectious disease.
