AI Predicts DNA Fixes: New Tool Edits Genes with Precision

A new method for editing DNA uses artificial intelligence to make changes with high accuracy. The technique, called AlphaFold3-based contact modelling, predicts how proteins interact with DNA. This allows scientists to target and alter a single “letter” in the genetic code without affecting nearby genes. Traditional gene editing tools, like CRISPR, can sometimes cut DNA in the wrong place. The new AI model helps avoid these errors by mapping the exact contact points between editing proteins and the DNA strand. Researchers say this approach reduces unintended mutations, making gene therapy safer. The process works by feeding the AI a protein’s structure and its target DNA sequence. The model then calculates the most likely binding site. This prediction guides the editing tool to the correct location, enabling a precise “base edit”—a change to one of the four chemical bases that make up DNA. While still in early stages, the method could eventually treat genetic diseases caused by a single typo in a patient’s genome. Experts caution that more testing is needed before human trials begin.