B1637
Title: Identification of differentially expressed genes via knockoff statistics in single-cell RNA sequencing data analysis
Authors: Lixia Yi - University of Pittsburgh (United States) [presenting]
Linxi Liu - University of Pittsburgh (United States)
Abstract: Single-cell RNA sequencing (scRNA-seq) is a high-throughput RNA sequencing technology that provides high-resolution gene expression data at the single-cell level. With scRNA-seq data, an important type of statistical analysis is to identify differentially expressed genes (DEGs) in case-control studies, as results from DEG analysis can contribute to a more comprehensive understanding of the disease mechanism and new discovery of potential risk factors. However, given the burden of multiple testing due to a large number of genes and low transcript capture rate in scRNA-seq data, DEG identification may suffer from low power, especially when sample size is limited. Co-expressed genes and unobserved confounders may also lead to an inflated Type-I error. A new method for DEG identification in scRNA-seq data analysis under the knockoff framework is introduced to overcome these difficulties. The method starts with imputing missing gene expressions by taking advantage of correlations among genes and then generates knockoff variables in a computationally efficient way. FDR control and power gain of the new method are illustrated on a range of synthetic and real data sets. The approach is also applied to single-cell transcriptomic analysis of Alzheimer's disease and cross-reference the discovered genes with other studies.
