Why AI struggles to decipher the complexity of human genomes

Scientists are grappling with the intricate nature of the human genome, which defies simple interpretation. While the Human Genome Project mapped the sequence of DNA in 1990-2003, it revealed that only 2% of the genome consists of genes. The rest, once considered 'junk DNA,' is now understood to play a critical role in gene regulation. AI models like Evo 2 and AlphaGenome are being used to analyze genomic data, but they struggle to capture the nuanced, context-dependent processes that govern gene expression. Biologists argue that the genome is not a straightforward code but a complex system shaped by evolution over 4 billion years. Karen Adelman of Harvard Medical School notes that gene regulation, involving noncoding DNA, is far more intricate than previously thought. Enhancers, which control gene activity, are numerous and often located far from the genes they influence, creating challenges in understanding their function. Wendy Bickmore of the University of Edinburgh highlights the difficulty in mapping enhancers and their interactions, emphasizing the dynamic, three-dimensional structure of chromatin as essential to gene regulation.