From the Dark Genome to Immunotherapy Targets: Artificial Intelligence Enables the Identification of Noncanonical Neoantigens

Emerging evidence has demonstrated that a previously defined dark genome might encode short peptides that bind human leukocyte antigen as noncanonical antigens to activate immunity. Systematic discovery and functional characterization of noncanonical neoantigens will promote the understanding of the human genome and offer profound insights into cancer immunotherapy. The continuous development of experimental and artificial intelligence techniques and the explosive growth of multimodal omics data have provided unprecedented opportunities and challenges for the identification of noncanonical neoantigens. In this review, we summarize our current understanding of various mechanisms that potentially generate noncanonical neoantigens and highlight the critical roles of previously unannotated peptides derived from noncoding regions in cancer biology. Moreover, we discuss recent technological advances and various types of artificial-intelligence-driven methods for identifying noncanonical neoantigens. Furthermore, we discuss the potential use of neoantigens derived from noncoding regions as therapeutic targets, shedding light on the function of the dark genome.