UC San Diego AI Decodes Gene On-Switch in 500,000 DNA Tests, Finding It in 60% of Genes
Updated
Updated · Mirage News · Aug 23
UC San Diego AI Decodes Gene On-Switch in 500,000 DNA Tests, Finding It in 60% of Genes
3 articles · Updated · Mirage News · Aug 23
Summary
Roughly 60% of human genes contain the initiator sequence after UC San Diego researchers trained an AI model to recognize the DNA pattern that marks where gene expression begins.
About 500,000 initiator variants were measured with high-throughput sequencing, giving the machine-learning system enough data to predict the sequence's presence or absence in human genes.
Those predictions could help researchers gauge how initiator mutations disrupt gene activity and contribute to disorders including cancer.
The team said the data and model may also aid design of synthetic promoters—DNA switches tailored to turn genes on or off for specific uses.
More broadly, the study points toward larger AI models that could decode the wider gene-expression code embedded across the human genome's 6 billion DNA bases.