Semi-synthetic Negative & Positive Control Enhancing the Reliability and Power in Single-cell and Spatial Omics Data Analysis

The presentation outlines the development and applications of advanced simulation tools designed to improve the reliability and interpretability of single-cell and spatial omics data analysis. Centered around the scDesign series (scDesign, scDesign2, scDesign3) and the newer scReadSim, scDesignPop, scIsoSim, and scEditSim, the work emphasizes three main goals: benchmarking computational methods, interpreting real data, and generating alternative datasets for robust statistical inference. scDesign3 simulates gene expression count data by modeling realistic joint and marginal distributions of molecular features such as genes, while scReadSim extends this to sequencing reads, capturing k-mer distributions and error rates. These tools provide controlled, realistic synthetic data that aid in method validation, interpretation of biological signals, and mitigation of false discoveries—exemplified by applications like ClusterDE, which addresses double-dipping in post-clustering differential expression analysis. The presentation ultimately showcases how tailored simulation frameworks can bridge the gap between statistical modeling and biological complexity.