AI helps take next stride on proteins in medicine
Researchers introduced ContactSeek, an AI-guided method that uses AlphaFold3 to improve DNA base editing specificity and reduce off-target effects.
- DNA base editing changes one DNA base at a target location, aiming to avoid full double-strand DNA breaks.
- Off-target effects are unintended DNA changes at other genomic sites; they can cause safety and accuracy problems in gene therapy.
- AlphaFold3 predicts how a protein is likely to interact with DNA, helping scientists focus on the protein parts most likely to contact DNA.
- Current DNA base editors can affect unintended DNA regions, because the protein parts that contact DNA are not always predicted well by existing 3D modeling.
What happened
Researchers introduced ContactSeek, an AI-guided method designed to improve DNA base editing for gene therapy. DNA base editing is described as promising for gene therapy, but base editors can affect unintended DNA regions, which makes precision and safety central requirements.
ContactSeek uses the AI structure-prediction model AlphaFold3 to estimate which parts of the editing protein are likely to contact DNA. Using these predictions, researchers design improved variants of common editors, including Cas9, with the reported goals of increasing targeting accuracy and reducing off-target effects.
UPSC can frame ContactSeek as an attempt to raise gene-editing precision. AlphaFold3-based prediction of protein–DNA contacts supports redesign of DNA base editors, aiming to reduce unintended DNA edits that are a key safety concern in gene therapy.