Systematic benchmarking of ambient RNA decontamination tools to advance precision in single-cell transcriptomic analysis

Single-cell RNA sequencing measures gene expression in individual cells, but RNA released from damaged cells can be captured alongside RNA from intact cells, producing spurious signals that affect cell identification and biological interpretation. Although many computational tools aim to remove ambient RNA, their relative performance has not been systematically evaluated. Using simulated and experimental datasets from diverse tissues and species, we assessed seven tools for accuracy in estimating contamination levels, robustness across biological and technical conditions, and sensitivity to subtype-specific contamination. Here we show that the tools have distinct strengths, underscoring the need to match methods to data characteristics and research goals. DecontX provides the most accurate contamination-level estimation. scAR is the most robust across conditions but tends to overestimate contamination, whereas CellClear performs best for resolving closely related cell subtypes. This benchmark guides method selection to improve the reliability of single-cell analyses and identifies priorities for future method development. Researchers present a study where they evaluate computational tools for removing ambient RNA from single-cell data based on their accuracy, robustness, and sensitivity to subtype-specific contamination, guiding method selection and advancing precision in single-cell research.