PlotSimilarityHeatmap: Create a comprehensive similarity heatmap plot

In davnovak/SingleBench: Framework for benchmarking and optimising dimension-reduction and clustering pipelines in expression data analysis

Description

Creates a comprehensive plot showing how the results of clustering map onto manually labelled populations using different matching strategies.

Usage

PlotSimilarityHeatmap(
  benchmark,
  idx.subpipeline,
  idx.n_param = NULL,
  idx.run = 1
)

Arguments

benchmark	object of type Benchmark, as generated by the constructor Benchmark and evaluated using Evaluate
idx.subpipeline	integer value: index of sub-pipeline that includes a clustering step
idx.n_param	integer: index of n-parameter iteration of interest. Default value is NULL
idx.run	integer: if clustering was run repeatedly for stability analysis, which run should be used to plot the heatmap. Default value is 1

Details

Using this, you can look at which cell populations were identified correctly or incorrectly by automated clustering and what kinds of mistakes the clustering set-up made. Moreover, the heatmap shows eventual differences between matching clusters to populations bijectively (one-to-one) and taking the best cluster for each population (and vice versa). A comparison of precision, recall and F1 scores for each match is also provided.

To create the plot, you need to specify which sub-pipeline and n-parameter iteration you want to look at. If you choose multiple n-parameter iterations, a list of plots is returned.

davnovak/SingleBench documentation built on Dec. 19, 2021, 9:10 p.m.