10 Set Intersections

This section provides an interface to create visualizations for intersections between differentially expressed genes identified in various contrasts within an experiment. Two types of plots are available: Venn Diagram for 2 to 5 sets (contrasts) and Upset Plot for up to 40 sets (contrasts). When more than 5 sets are selected the Venn diagram plot will display an error message. As a matter of personal preference we recommend using Venn Diagram for up to 4 intersecting sets and Upset plot for a larger number of sets. As both visualization methods use the same inputs they also share the same set of controls.

10.1 Set Intersection Control Panel

Select Contrast: Allows users to select desired contrasts (comparisons) that were used during differential expression analysis to perform set intersection (default: All).

Select Gene Subsets: By default, all differentially expressed genes (both up- and down-regulated) will be used for the set intersection. Selecting Upregulated or Downregulated will limit intersecting genes to either only up- or downregulated regulated genes.

FDR or P-value:, FDR/P-value Cutoff: and Log Fold Change: introduce statistical and effect size cutoffs that define which genes are deemed to be differentially expressed. The default values for these parameters are FDR, 0.05 and 1 respectively. We strongly discourage using uncorrected P-values as it greatly increases the number of false-positives among identified differentially expressed genes and complicates interpretation.

10.2 Venn Diagram Tab

The Venn diagram plot creates a proportional (when possible) Venn diagram (Figure 10.1). Each circle or ellipse corresponds to a set of differentially expressed genes (up-, down-, or both) given the specific statistical constraints. The name of each contrast and the number of genes in each subset is shown on the plot. Both the plot and an associated subset table can be downloaded. On the backend rnaseqDRaMA uses custom R functions to calculate multiset intersection and heavily modified venn.diagram() function from R venn.diagram package (H. Chen 2018).

Non-proportional Venn Diagram of 3 sets

Figure 10.1: Non-proportional Venn Diagram of 3 sets

10.3 UpSet Plot Tab

UpSet plot is a recently introduced visualization technique (Lex et al. 2014) for set analysis that uses a bargraph to indicate the size of each intersecting element and a combination matrix underneath the bargraph to indicate the relationships between subsets (Figure 10.2).

Similarity between upset and Venn Diagram

Figure 10.2: Similarity between upset and Venn Diagram

Figure 10.2 shows the correspondence between a three-set Venn diagram and an UpSet plot. In this representation each row of the lower combination matrix correspond to a primary set while the column corresponds to intersection subsets. Dark blue dots indicate that a subset is a part of an intersection. First three columns have only a single dot corresponding to a primary set. These columns correspond to the unique elements of primary sets that do not participate in any intersection. The height of the matching bars in the upper bar graph reflect the number of elements in each subset.

For example, among differentially expressed genes in FWTFAVC-vs-FWTFoB contrast 1743 genes (the left-most bar) are unique and do not participate in any intersections. Conversely, the subset with 188 genes represent by the rightmost bar has dark blue dot in each row, indicating that genes in this subset were differentially expressed in all three contrasts (see the central subset in the Venn diagram).

References

Chen, Hanbo. 2018. “VennDiagram: Generate High-Resolution Venn and Euler Plots.” https://CRAN.R-project.org/package=VennDiagram.

Lex, A., N. Gehlenborg, H. Strobelt, R. Vuillemot, and H. Pfister. 2014. “UpSet: Visualization of Intersecting Sets.” IEEE Transactions on Visualization and Computer Graphics 20 (12): 1983–92. doi:10.1109/TVCG.2014.2346248.