Summary
Transitions in cell identity are fundamental to development, reprogramming, and disease. Single-cell technologies enable the dissection of tissue composition on a cell-by-cell basis in complex biological systems. However, highly-sparse single-cell RNA-seq data poses challenges for cell-type identification algorithms based on bulk RNA-seq. Single-cell analytical tools are also limited, where they require prior biological knowledge and typically classify cells in a discrete, categorical manner. Here, we present a computational tool, ‘Capybara,’ designed to measure cell identity as a continuum, at single-cell resolution. This approach enables the classification of discrete cell entities but also identifies cells harboring multiple identities, supporting a metric to quantify cell fate transition dynamics. We benchmark the performance of Capybara against other existing classifiers and demonstrate its efficacy to annotate cells and identify critical transitions within a well-characterized differentiation hierarchy, hematopoiesis. Our application of Capybara to a range of reprogramming strategies reveals previously uncharacterized regional patterning and identifies a putative in vivo correlate for an engineered cell type that has, to date, remained undefined. These findings prioritize interventions to increase the efficiency and fidelity of cell engineering strategies, showcasing the utility of Capybara to dissect cell identity and fate transitions. Capybara code and documentation are available at https://github.com/morris-lab/Capybara.
