Covarying neighborhood analysis identifies cell populations associated with phenotypes of interest from single-cell transcriptomics

As single-cell datasets grow in sample size, there is a critical need to characterize cell states that vary across samples and associate with sample attributes like clinical phenotypes. Current statistical approaches typically map cells to clusters then assess differences in cluster abundance. We present covarying neighborhood analysis (CNA), an unbiased method to identify associated cell populations with greater flexibility than cluster-based approaches. CNA characterizes dominant axes of variation across samples by identifying groups of small regions in transcriptional space—termed neighborhoods—that covary in abundance across samples, suggesting shared function or regulation. CNA performs statistical testing for associations between any sample-level attribute and the abundances of these covarying neighborhood groups. Simulations show that CNA enables more sensitive and accurate identification of disease-associated cell states than a cluster-based approach. When applied to published datasets, CNA captures a Notch activation signature in rheumatoid arthritis, identifies monocyte populations expanded in sepsis, and identifies a novel T-cell population associated with progression to active tuberculosis.