This video introduces the basic workflow for using the Basic Plug-in in Subio Platform to visually inspect RNA-Seq data and compare the results of differential expression analysis.
The dataset used in this video is the same GSE49110 dataset used in the RNA-Seq Data Analysis Tutorial. To compare three siRNA treatments, normalized Gene Counts have been converted into Log2 Ratios against the control group siC. Using this data, the video demonstrates filtering, PCA, hierarchical clustering, heatmaps, DEG analysis against the control group siC, and comparison of DEG lists using Venn diagrams.
To keep the operation flow easy to follow, this video omits detailed explanations of normalization and preprocessing. In practice, the normalized Gene Counts have been preprocessed using steps such as Low Signal Cutoff and Fill Missing Values before being converted into Log2 Ratios against the control group. For detailed procedures, please see the RNA-Seq Data Analysis Tutorial.
Subio Platform Basic Plug-in Quick Tutorial: Filtering, PCA, Heatmap, DEG Analysis, and Venn Diagram
Analysis workflow using the Basic Plug-in
In this video, the analysis proceeds through the following steps.
- Exclude genes with unstable measurements in the low-count region
- Exclude genes that show almost no expression change
- Use PCA to check the relationships among samples
- Use hierarchical clustering and heatmaps to examine expression patterns
- Extract DEG lists against the control group siC
- Save Up and Down lists separately
- Compare multiple DEG lists using Venn diagrams
- Examine the union list of DEGs using a heatmap
Filtering should be decided while looking at the data
In RNA-Seq data, measurements in the low-count region tend to be unstable. If they are used directly for PCA, clustering, or differential expression analysis, the results can become difficult to interpret.
In this video, the threshold is not applied mechanically as a fixed value. Instead, the region to exclude is decided while looking at scatter plots that visualize the Gene Counts.
Genes that show almost no expression change do not always need to be excluded for DEG analysis. However, they are important filtering targets for clustering. If a large number of genes with no meaningful expression change are included in clustering, the expression patterns that you really want to see can become harder to interpret.
Interpret PCA and heatmaps together
PCA allows you to check the relationships among samples and the overall differences in expression profiles. In this dataset, the two control samples are located near the origin, and the siRNA-treated samples mainly move toward the lower-right direction.
Hierarchical clustering and heatmaps then allow you to examine which groups of genes are up-regulated or down-regulated in each treatment group. PCA and heatmaps should not be interpreted separately. It is important to interpret them together, so that you can understand what kinds of expression patterns explain the sample-level differences observed in the PCA plot.
Extract DEG lists and compare them with Venn diagrams
Using the Compare 2 Groups tool in the Basic Plug-in, you can extract genes that are up-regulated or down-regulated relative to the control group.
In this video, siC is used as the reference group, and DEG lists are extracted for siE1, siE2, and siE3. The Up and Down lists are saved separately.
When extracting DEG lists, the threshold is decided not only by checking the Volcano Plot, but also by using a scatter plot that shows the relationship between average expression level and expression change.
The video also uses the Venn diagram tool, a basic function of Subio Platform, to compare the overlaps and differences among multiple DEG lists. In this dataset, the up-regulated DEG lists contain a relatively large number of common genes, whereas the down-regulated DEG lists contain more group-specific genes.
Do not rely only on automated analysis. Check the data visually.
The purpose of this video is not only to show how to operate the Basic Plug-in. In RNA-Seq data analysis, it is important not only to obtain results automatically, but also to visualize the data at each step and check the validity of filtering, the relationships among samples, expression patterns, and the similarities and differences among DEG lists.
Subio Platform allows you to visually inspect RNA-Seq data, check analysis results, adjust conditions when necessary, and proceed toward biological interpretation.