DeepMind released AlphaGenome, a large DNA‑sequence model reported in Nature that predicts regulatory effects of noncoding variants across up to one million base pairs. The model outputs thousands of molecular properties—gene starts and ends, splicing junctions, RNA production, and chromatin accessibility—enabling rapid scoring of variant impacts for research and target discovery. AlphaGenome’s long‑context architecture addresses distal regulatory interactions and runs variant impact scoring in roughly one second per variant, positioning it as a tool for rare variant interpretation, therapeutic target identification, and synthetic regulatory DNA design. The team highlights applications in splicing‑related diseases and genetic mechanism elucidation. For genomics R&D, AlphaGenome advances computational annotation of the regulatory genome and could accelerate candidate selection and functional interpretation, though experimental validation will remain essential.