ai models struggle to crack genome code
researchers find ai isn't ready to tackle complex genomes, despite models like evo 2 and alphagenome, due to messy 3d dna structure.
researchers are using ai models like evo 2, genos, and google deepmind's alphagenome to make predictions about how differences in dna sequence affect biological processes. however, experts say that these models may not provide a real understanding of how genomes work due to the complex and dynamic nature of dna. the human genome has only been partially deciphered, with 2% consisting of actual genes, and gene regulation is still not well understood. biologists like karen adelman argue that ai black boxes may not be enough to crack the genome code.