Card deck

NTv3: one model to read, predict and design DNA

Nucleotide Transformer v3 is a multi-species model for DNA, trained on 9 trillion letters and reading up to 1 million bases at once. One backbone predicts gene activity and genome structure at single-base resolution, and can be adapted to design regulatory DNA with a chosen activity level.

Genomics × AIPaper

1 / 11 NTv3: one model to read, predict and design DNA

An eleven-slide breakdown of Nucleotide Transformer v3: how a U-Net backbone reaches megabase context with far fewer parameters, learns by filling in masked DNA, sets state of the art across 106 benchmark tasks, and turns generative to design enhancers that were validated in living cells.

Read the paper: A foundational model for joint sequence-function multi-species modeling at scale for long-range genomic prediction (bioRxiv, 2025).