Analysis and visualization of RNA-Seq expression data using RStudio, Bioconductor, and Integrated Genome Browser

Abstract

Sequencing costs are falling, but the cost of data analysis remains high, often because unforeseen problems arise, such as insufficient depth of sequencing or batch effects. Experimenting with data analysis methods during the planning phase of an experiment can reveal unanticipated problems and build valuable bioinformatics expertise in the organism or process being studied. This protocol describes using R Markdown and RStudio, user-friendly tools for statistical analysis and reproducible research in bioinformatics, to analyze and document the analysis of an example RNA-Seq data set from tomato pollen undergoing chronic heat stress. Also, we show how to use Integrated Genome Browser to visualize read coverage graphs for differentially expressed genes. Applying the protocol described here and using the provided data sets represent a useful first step toward building RNA-Seq data analysis expertise in a research group.

Figure 1. Example code chunk displayed in RStudio

Embedded R code (called a “chunk”) is set apart from the rest of the Markdown document using triple back tics, followed by the letter “r” in curly braces. The Knit HTML button or the “Chunks” menu can be used to run all or part of the chunks in a Markdown document. Running “Knit HTML” Optional formatting options (such as figure width and height) can also appear between the curly braces.

Figure 2. MDS and clustering plots showing relationships between sample types

(A) Multidimensional scaling (MDS) plot. Distance between sample labels indicates similarity. (B) Cluster dendrogram. Number of branches separating samples indicates similarity. Treatment samples are named T1 through T5. Control samples are named C1 through C5.

Figure 3. Plot showing relationship between average expression and fold-change

Horizontal lines indicate fold-change of two. Red indicates genes called as differentially expressed at FDR of 0.005 or smaller.

Figure 4. Integrated Genome Browser showing pollen data sets and gene models

(A) Whole chromosome view with coverage graphs and gene models from the ITAG2.4 annotations release. Coverage graphs are from treatment (T3, T5) and control (C1, C2) samples. The arrow indicates the location of the most highly expressed gene on the chromosome. (B) Zoomed-in view of the highly expressed gene indicated in (A). The gene encodes a putative polygalacturonase.

Figure 5. Zooming in on a differentially expressed gene

(A) Coverage graphs for a differentially expressed (DE) gene. For comparison, a nearby gene that is not differentially expressed is also shown. (B) Right-click context menu. Right-clicking a gene model triggers options to do a BLAST search, view the sequence in a new window, or search Google. (C) Image export. Selecting File > Export Image saves an image file with the current view within IGB.