DeepMind's AlphaGenome 2 cracks DNA's dark side, no pipettes needed
Google DeepMind in London announced AlphaGenome 2, an update to its AI tool that predicts the molecular effects of non-coding DNA, such as on the expression of other genes. Non-coding sequences make up 98% of the human genome and are among the hardest to decipher. The new model can predict which cell types a DNA variant is likely to act in. It was trained on data from more than 100 million individual cells across 68 tissues, whereas the first version used bulk samples of mixed cells.
Why it matters: By predicting which cell types a DNA variant affects within the 98% of the genome that doesn't code for proteins, this tool could help researchers understand how genetic differences influence disease without lab experiments.
Comments
No comments yet.