AI and single-cell genomics advance biological research, says MSK biologist

Dr. Christina Leslie, a computational biologist at Memorial Sloan Kettering Cancer Center (MSK), is utilizing artificial intelligence to interpret complex biological data [93]. Her research focuses on the regulation of gene expression in mammalian cells, specifically within the immune system and in cancer [93].

Leslie’s work integrates three main types of data: single-cell genomics, spatial transcriptomics, and 3D genomics [93]. Unlike traditional "bulk" analysis, which examines the combined molecular data of a tissue sample, single-cell genomics allows for the study of the genetic activity and epigenetic state of individual cells [93]. Spatial transcriptomics provides information on the physical location of those cells within a tissue, while 3D genomics maps the physical folding and organization of DNA inside the cell nucleus [93].

While AI provides the computational capacity to analyze massive datasets—some reaching 150 million cells—Leslie notes that current models face limitations [93]. She observes that many models adapted from natural language processing or computer vision are too generic and do not fully represent the biological structure of gene regulation [93]. Consequently, the ability of these models to perform automated cell labeling remains too imprecise for many biological tasks [93].

These technologies have facilitated new biological discoveries [93]. Researchers at MSK used single-cell methods to identify "Thetis cells," a specialized immune cell type that may explain how the body develops tolerance to food and gut bacteria [93]. Additionally, studies on colorectal cancer revealed two distinct types of regulatory T cells—IL-10 positive and IL-10 negative—that play opposing roles in tumor growth and restraint [93]. Leslie suggests that advancing the field will require closer collaboration between computational researchers and bench scientists [93].