Artificial Intelligence in Genomics: Transforming the Future of Biological Discovery: A Review

Abstract

Artificial intelligence has become one of the most consequential methodological developments in contemporary genomics because it offers ways to learn complex biological patterns from genome sequences, population-scale variation catalogues, functional genomic assays, single-cell profiles, epigenomic maps, protein-structure resources and clinical molecular datasets. This review examines how artificial intelligence is transforming biological discovery across the genomic sciences. It considers the progression from classical machine learning to deep learning, transformer-based foundation models, graph-based integration and generative modelling, with particular emphasis on sequence interpretation, variant calling, functional annotation, regulatory genomics, protein structure prediction, single-cell analysis, multi-omics integration, genome editing and genomic medicine. The review argues that artificial intelligence is changing genomics from a mainly descriptive and catalogue-building discipline into a more predictive, integrative and hypothesis-generating science. However, it also stresses that artificial intelligence cannot substitute for biological reasoning, experimental validation or clinically governed interpretation. Major challenges include limited interpretability, technical artefacts, data leakage, ancestry bias, underrepresentation of many global populations, privacy risk, reproducibility concerns and the difficulty of translating benchmark performance into clinical utility. Future progress will depend on more diverse genomic datasets, transparent benchmarking, pangenome-aware modelling, biologically interpretable architectures, stronger links between prediction and perturbation experiments, and governance systems that protect individuals and communities while enabling responsible data sharing. Properly used, artificial intelligence can accelerate discovery across molecular biology, disease genetics and precision medicine, but its value will depend on how rigorously predictions are validated and how fairly genomic benefits are distributed.