Artificial intelligence (AI) technologies have recently undergone transformative growth in capabilities. In human genetics, AI is rapidly advancing our ability to reveal the effects of genetic variation. This review explores recent progress and remaining challenges across the diverse applications of AI in genotype-to-phenotype mapping, from predicting the functional and clinical consequences of mutations, to identifying causal genes, to estimating disease risk. Particular emphasis is placed on the growing utility of general-purpose foundation models trained on massive genomic data, including DNA and protein language models, alongside areas where narrower machine-learning approaches still dominate. The review concludes with key considerations for future progress and impact.
Nadav Brandes (Wed,) studied this question.