China releases free DNA-screening AI to speed rare disease diagnosis

Rare genetic diseases have long posed a diagnostic challenge for patients and physicians. Because these conditions are mostly genetic in origin, identifying the underlying mutation requires not only sequencing the patient's DNA but also correctly interpreting what that mutation means for their health and treatment options. Patients often spend years moving between specialists and undergoing inconclusive tests, a grueling process the medical community calls the "diagnostic odyssey," while the answer they seek remains hidden within three billion base pairs of DNA in the human genome.

To accelerate this process, Chinese researchers at BGI-Research have created OneGenome, an open-source AI system designed to interpret DNA sequences and connect them to clinical knowledge. Unlike simpler tools that attempt to read DNA sequences directly, OneGenome is trained to reason about the clinical consequences of genetic mutations using accumulated human medical logic. The system was built by integrating two components: Genos, a genomic foundational model released in the previous year that was trained on human genomes to represent diverse global populations, and large language model capabilities that allow it to reason across medical literature and clinical knowledge.

According to BGI-Research, OneGenome has demonstrated strong performance in rigorous testing. In multiple clinical diagnostic and medication guidance tests, it has outperformed general large language models such as DeepSeek-v4 as well as other traditional gene-specific models. This success suggests that combining specialized genomic training with general medical reasoning capabilities yields better results than either approach alone. Because the tool is released open-source and free to use, it is being made available to researchers, clinicians, and institutions worldwide, potentially reducing the diagnostic burden for patients navigating rare genetic diseases.