DeepMind released new AI models designed to interpret regulatory and splicing‑related DNA variation across the noncoding genome. The Alphagenome/AlphaGenome suite applies advanced embeddings and predictive architectures to map variant effects on gene expression and splicing, addressing a major interpretation gap in whole‑genome analysis. Researchers say these tools may accelerate target discovery and variant prioritization in both rare‑disease diagnostics and complex trait studies; the models aim to translate noncoding variation into actionable hypotheses for lab validation and therapeutic development.