The ChatGTP For Genomes

This piece, by Onno Berkan, was published on 02/11/25. The original text, by Nguyen et al., was published by Science on 11/15/24.

Researchers at Stanford have developed Evo, a new AI model designed to understand and work with genetic information at multiple biological scales. Trained using prokaryotic genomes, this 7-billion-parameter model can analyze and generate DNA sequences ranging from individual molecules to entire (still prokaryotic) genomes, advancing our ability to understand and manipulate biological systems significantly.

What makes Evo particularly special is its ability to work across different types of biological information. Due to their dependent nature, it can understand DNA, RNA, and proteins simultaneously, rather than being limited to just one type like previous models. The model was trained on an enormous dataset of 2.7 million bacterial and viral genomes, giving it a comprehensive understanding of how genetic information works in these organisms.

Another one of Evo's impressive abilities is generating and analyzing very long DNA sequences– up to more than 1 million base pairs in length. This is particularly noteworthy because when Evo generates these long sequences, they show characteristics similar to real bacterial genomes, meaning they could be used (researchers found that one in eleven gene sequences produced by the model work, which is still a significant improvement over current methods.) 

The model also accurately predicts how genetic changes might affect an organism's survival and function. It can understand the complex relationships between different parts of a genome and how they work together, something that previous models couldn't achieve. This makes it especially valuable for understanding how genetic modifications might impact living organisms.

Evo is not perfect, however, as (just like any other generative AI), it makes mistakes. Not all of its creations work. But, as its creator argues adamantly, it’s much better than any human out there.

Want to submit a piece? Or trying to write a piece and struggling? Check out the guides here!

Thank you for reading. Reminder: Byte Sized is open to everyone! Feel free to submit your piece. Please read the guides first though.

All submissions to berkan@usc.edu with the header “Byte Sized Submission” in Word Doc format please. Thank you!