Skip to content

Step 3: Results Visualisation

Jump to bottom
Shreya Johri edited this page Apr 10, 2023 · 1 revision

BEANIE provides several useful visualization plots to explore the results of the analysis. These plots include the Bar Plot, Dropout Plot, Gene Rank Heatmap, and Upset Plots.

Bar Plot

All gene signatures which have statistically-significant empirical p-values are shown in this plot. The robustness of the gene signature to sample exclusion, and enrichment directionality are also shown in the plot. To generate this plot -

bobj.BarPlot()

Dropout Plot

This plot is especially for visualizing the non-robust signatures in detail. It shows three key features -

  1. Which samples led to the non-robustness of a particular gene signature (center heatmap)
  2. Each sample’s individual contribution to non-robustness of gene signatures (right bar-plot)
  3. Spread of the Fold Rejection Ratio (FRR) for every gene signature, corresponding to each sample’s exclusion (top box-plot). To generate this plot -
bobj.SampleDropoutPlot()

Gene Rank Heatmap

This plot shows the top-ranked genes (determined according to bobj.RankGenes()) for each gene signature. By default, the plot shows the 5 gene signatures with the lowest empirical p-values, but it is recommended to use this plot to visualize other gene signatures too. A list of gene signature names to be plotted can be passed as a parameter to this function.

# visualize the top 5 gene signatures with the lowest empirical p-values
bobj.GeneRankHeatmap()

# visualize other gene signatures
bobj.GeneRankHeatmap(signature_names = ["signature_1", "signature_2", "signature_3"])

Upset Plots

This plot is used for visualizing the overlap between the top 10 ranked genes across different gene signatures. A list of gene signature names can be provided. If no list is provided, the top 5 signatures with the lowest empirical p-values are used by default.

bobj.GeneRankUpsetPlot(signature_names=["signature_1", "signature_2", "signature_3", "signature_4", "signature_5"])

Similarly, the overlap between all genes of the gene signatures can also be seen. A list of gene signature names can be provided. If no list is provided, the top 5 signatures with the lowest empirical p-values are used by default.

bobj.SignatureGenesUpsetPlot(signature_names=["signature_1", "signature_2", "signature_3", "signature_4", "signature_5"])
Clone this wiki locally