scDIG software study enables interactive single-cell population gating

The open-access scDIG application combines feature selection, feature-weighted kernel density estimation and interactive contour-based gating to help researchers define cell populations in two-dimensional projections of single-cell RNA and protein data. Designed for scRNA-seq and CITE-seq datasets, the R Shiny tool offers an interactive middle ground between subjective manual selection and fully automated clustering algorithms that may obscure rare but biologically significant populations. Delineating meaningful populations in these views can be difficult to reproduce when decisions rely mainly on an analyst’s visual judgment.

The workflow begins with a bimodal index-driven procedure for selecting informative features. It then uses feature-weighted kernel density estimation and displays the resulting density as interactive contours that can be gated. The application was built in R Shiny and supports both gene-expression measurements and combined transcript-protein data, allowing the same interactive framework to be used for scRNA-seq and CITE-seq datasets.

The reported aim is to combine the interpretability of expert-guided population selection with a systematic basis for defining and reproducing those selections. In a software application study using CITE-seq peripheral blood mononuclear cell data from participants in the Cardiovascular Assessment Virginia cohort, scDIG resolved transcriptionally distinct CD4-positive T-cell subpopulations within continuous embeddings that conventional clustering did not readily capture. Applied to RNA-expression data, the tool also identified GZMB-enriched cytotoxic lymphocyte subsets.

The authors also conducted sensitivity analyses to assess whether the gates identified in two-dimensional views corresponded to coherent populations in the original high-dimensional feature space. They reported that the defined gates corresponded to coherent cellular groupings, supporting the use of interactive density contours as a way to connect visual population definition with underlying measurements. scDIG is freely available for non-commercial use, and its source code is available through GitLab.

Its central contribution is an interactive method for defining cell populations in single-cell embeddings while retaining researcher oversight. By translating weighted density estimates into adjustable contours, the application provides a practical framework for investigating distinct cell groups in scRNA-seq and CITE-seq data and for evaluating whether selected gates remain coherent in high-dimensional feature space.

Subject of Research: Interactive density-based gating of single-cell proteomic and transcriptomic data

Article Title: scDIG: an R Shiny application for interactive density-based gating of single-cell proteomic and transcriptomic data

Article References: Bombina, P., Bellapu, A., Fogel, L., Coombes, K. R., & Ley, K. (2026). scDIG: an R Shiny application for interactive density-based gating of single-cell proteomic and transcriptomic data. BMC Bioinformatics. https://doi.org/10.1186/s12859-026-06718-3

Image Credits: AI Generated

DOI: 10.1186/s12859-026-06718-3

Keywords: scDIG, single-cell analysis, scRNA-seq, CITE-seq, cell gating, kernel density estimation, R Shiny, bioinformatics, cytotoxic lymphocytes, software, Shiny, application