Simulating Longitudinal Single-cell RNA Sequencing Data with RESCUE

Abstract

As single-cell RNA-sequencing (scRNA-seq) becomes more widely used in transcriptomic research, complex experimental designs, such as longitudinal studies, become increasingly feasible. Longitudinal scRNA-seq enables the study of transcriptomic changes over time within specific cell types, yet guidance on analytical approaches and resources for study planning, such as power analysis, remains limited. Data simulation is a valuable tool for evaluating analysis method performance and informing study design decisions, including sample size selection. Currently, most scRNA-seq simulation methods simulate cells for a single sample, thus ignoring the between-sample and between-subject variability inherent to longitudinal scRNA-seq data. Here, we introduce RESCUE (REpeated measures Single Cell RNA-seqUEncing data simulation), a novel method that simulates longitudinal scRNA-seq data using a gamma-Poisson frame-work and incorporates additional variability between samples and subjects. We demonstrate our methods ability to reproduce important data properties and demonstrate its application in study planning. RES-CUE is implemented as an R package and is available at https://github.com/ewynn610/RESCUE.