A Two-Part Mixed Model for Differential Expression Analysis in Single-Cell High-Throughput Gene Expression Data

Abstract

The high-throughput gene expression data generated from recent single-cell RNA sequencing (scRNA-seq) and parallel single-cell reverse transcription quantitative real-time PCR (scRT-qPCR) technologies enable biologists to study the function of transcriptome at the level of individual cells. Compared with bulk RNA-seq and RT-qPCR gene expression data, single-cell data show notable distinct features, including excessive zero expression values, high variability, and clustered design. We propose to model single-cell high-throughput gene expression data using a two-part mixed model, which not only adequately accounts for the aforementioned features of single-cell expression data but also provides the flexibility of adjusting for covariates. An efficient computational algorithm, automatic differentiation, is used for estimating the model parameters. Compared with existing methods, our approach shows improved power for detecting differential expressed genes in single-cell high-throughput gene expression data.