A0715
Title: Differential inference for single-cell RNA-sequencing data
Authors: Fangda Song - The Chinese University of Hong Kong, Shenzhen (China)
Kevin Y Yip - Sanford Burnham Prebys Medical Discovery Institute (United States)
Yingying Wei - The Chinese University of Hong Kong (Hong Kong)
Yingying Wei - The Chinese University of Hong Kong (Hong Kong) [presenting]
Abstract: Single-cell RNA-seq (scRNA-seq) experiments are becoming more and more complicated with multiple treatment or biological conditions. However, guidelines on experimental designs and rigorous statistical methods for a comparative scRNA-seq study with data collected from multiple conditions are still lacking. Existing multi-stage approaches to identifying differential cell-type abundance and differentially expressed genes between conditions suffer from high error rates because multi-stage approaches ignore uncertainties in previous stages and propagate errors in earlier stages to later stages. DIFseq, a Bayesian hierarchical model, is introduced to rigorously quantify the condition effects on both cellular compositions and cell-type-specific gene expression levels for scRNA-seq data. DIFseq substantially outperforms state-of-the-art methods in terms of the accuracy of cell type clustering, differential abundance, and differential expression inference for both simulated and real data. Moreover, to the best of knowledge, the conditions are derived for a valid design for a comparative scRNA-seq study for the first time.
