RNA-seq: Basic Bioinformatics Analysis

Abstract

Quantitative analysis of gene expression is crucial for understanding the molecular mechanisms underlying genome regulation. RNA-seq is a powerful platform for comprehensive investigation of the transcriptome. In this unit, we present a general bioinformatics workflow for the quantitative analysis of RNA-seq data and describe a few current publicly available computational tools applicable at various steps of this workflow. These tools comprise a pipeline for quality assessment and quantitation of RNA-seq data that starts from raw sequencing files and is focused on the identification and analysis of genes that are differentially expressed between biological conditions. © 2018 by John Wiley & Sons, Inc.

Keywords: RNA-seq; bioinformatics; differentially expressed genes; quantitative analysis of gene expression.

Figure 1.

Example of MDS plot of six RNA-seq samples. Each sample is shown as a two-dimensional point represented by text (sample name) and colored by group (condition). In this particular case, the group of three biological replicates from mutant samples (mut, shown in red) is well separated from the group of three biological replicates from wild-type samples (WT, shown in black). In the WT group, replicate 3 (WT.rep3) is separated from other replicates of the same group and may be a potential outlier.

Figure 2.

Example of a Volcano plot. Each gene is represented as a point in the space of absolute difference in expression value between two compared groups of replicates (log2 of fold change, logFC) as the x axis and the statistical significance of this difference (-log10 of FDR or P-value) as the y-axis. The plot has a characteristic shape reflecting a general relationship between fold change and statistical significance. The overall distribution of points is usually symmetrical between up-regulated genes (points to the right of x=0) and down-regulated genes (points to the left of x=0), with the majority of points located near the origin, which corresponds to small and statistically insignificant differences. Differentially expressed genes (highlighted in red) are defined here by the cutoffs of 2-fold change (logFC > 1 or logFC < −1) and statistical significance < 0.01.

Figure 3.

Heatmap of expression values of differentially expressed genes across individual samples. Clustering of expression patterns of samples (columns) and genes (rows) is represented by the dendrograms on top and on the left, respectively. Color indicates expression value (log10 of RPKM).