Evo 2

Headlines · 1

1. 2d
   Quanta essay asks whether AI can truly model the genome's complexity

   In an essay published by Quanta Magazine and reprinted by WIRED AI, the argument is that only about 2 percent of the human genome consists of protein-coding genes, and that understanding the genome now hinges less on what each gene does than on how genes are regulated, a question that may involve much of the rest of the genome. The piece cites genomic foundation models such as Evo 2, Genos and Google DeepMind's AlphaGenome, which are trained on large genomic datasets to predict how DNA sequence differences affect biological processes and ultimately traits including disease risk. The author argues that such computational black boxes may be useful but cannot deliver real understanding of how the genome works, since layers of regulation such as 3D chromatin folding and epigenetic marks suggest the relevant information is not contained in the sequence alone, and the genome may not fit a straightforward input-output modeling assumption.

   WIRED AI · 🔥 0

Experience and discussion from the community

Share my Evo 2 experienceAsk about Evo 2