Researchers at the Broad Institute and Mass General Brigham have developed a generative AI model designed to create short DNA segments called cis-regulatory elements (CREs). These CREs are critical components of the human genome that regulate gene activity by turning genes on or off in specific cell types. Using this AI model, the team generated over 5,800 synthetic CREs that were experimentally validated to maintain their intended gene regulatory functions in targeted cells. This breakthrough holds significant promise for gene therapies, as these synthetic DNA sequences could enable highly precise control over gene expression, potentially improving treatment approaches for a variety of diseases.
The AI model was trained on extensive DNA accessibility data across different cell types, allowing it to design CREs with cell-type-specific activity. This advancement merges machine learning with genomics to create tools that could modulate genes more accurately than traditional methods, marking a substantial step forward in therapeutic genetic engineering. The technology may eventually empower therapies that customize gene expression patterns, opening new avenues for treating complex diseases with fewer off-target effects.
This work reflects growing efforts within the biotech field to harness AI-driven approaches for biomedical innovation, combining computational power with synthetic biology to address unmet medical needs in gene regulation and therapy development.
