RNA sequencing analysis of neural cell lines: Impact of normalization and technical replication

Abstract

RNA sequencing offers a versatile platform for profiling biological samples. The novelty and flexibility of this technology has the potential to introduce technical variation at distinct points in the experimental workflow. We evaluated variation in RNA sequencing data acquired from commercially available cell lines cultured in our laboratory: human neural stem cells and normal human astrocytes. After normalizing data with three different methods, we used principal variance component analysis to estimate the contribution to technical variance from replicate cell lots, library preparations, and flow cells. Differentially expressed genes were evaluated using ANOVA analysis. Results indicate that the largest component of technical variance was library preparation. Moreover, comparative analysis of RNA sequencing data from the two cell types showed that the identification of differentially expressed genes and the contributions to variance are strongly influenced by the normalization method. Our results underscore the necessity for technical replication in RNA-seq experiments.