Many human diseases start out with single cells, which establish the problem by spreading out. Spatial transcriptomics, a groundbreaking method, allows scientists to take advantage of that activity, measuring all the gene activity in a tissue sample and mapping where the activity is occurring. Because it can potentially enable us to characterize molecular maps of cells within tissues — in effect, to map the baseline development of diseases such as cancer and Alzheimer’s — Nature Methods crowned it “Method of the Year” in 2020. 

Yet researchers confront a key, persistent bottleneck as they look to realize the translational potential of spatial transcriptomics: They lack the analytical tools to extract knowledge from the information they see. This is particularly true in the cancer microenvironment, which is highly complex. It comprises diverse cell types, which exhibit unique gene expression patterns that can vary over time. 

Enter the technique that biostatistician Mingyao Li, PhD, has developed. Dr.Li’s technique enables us to characterize the tumor microenvironment in detail. Findings from these analyses can be key to understanding tumor development, determining tumor malignancy, and identifying new therapeutic targets to enhance the effectiveness of cancer immunotherapy. Dr. Li and colleagues recently discussed their novel machine-learning-based technique in Nature Methods. 

Dr. Li’s technique will also apply to brain-related diseases such as Alzheimer’s disease (AD). In a brain affected with AD, beta-amyloid proteins gather in abnormal numbers and clump together to form plaques that collect between the neurons and disrupt cell function. Although we know that these complex alterations occur, we know little about the molecular changes and interactions that characterize this response. Dr. Li’s technique will fill this gap: It will allow us to spatially characterize the cell types around amyloid plaques and the construction of the cellular network around the pathogenic amyloid plaques. Findings from these analyses can potentially untangle the mechanisms behind the progression of AD and improve patient outcomes in the future. 

This technology promises nothing less than the transformation of diagnostic and prognostic medicine. “I anticipate that in the foreseeable future, spatial transcriptomics will be adopted by typical research labs, and clinics will use it for disease diagnosis and to guide personalized disease treatment,” says Dr. Li. “Having powerful analytical tools is essential to that progress,” she adds, “making this an exciting time for statisticians to tackle these problems and contribute to disease diagnosis and treatment.”