Online Supplement of Comparison of Transformations for Single-Cell RNA-Seq Data

This repository contains the code to generate the figures of our paper 'Comparison of Transformations for Single-Cell RNA-Seq Data' in Nature Methods https://www.nature.com/articles/s41592-023-01814-1.

The accompanying R package transformGamPoi is available via Bioconductor or at https://github.com/const-ae/transformGamPoi.

At https://shiny-portal.embl.de/shinyapps/app/08_single-cell_transformation_benchmark, we provide an interactive website to explore the benchmark results. The underlying source code is available at https://github.com/const-ae/transformGamPoi-ShinyApp.

Repository content

The repository is organized into four folders. The root directory contains some additional files for documentation and data plumbing purposes.

• benchmark contains the infrastructure for our benchmark of 22 different transformation methods. The job_overview.yaml is a declarative specification of all the parameter combinations (more than 58,000) that we benchmarked. The renv.lock file contains detailed information about each R package that was used in the benchmark using the renv system.
• notebooks contains the Rmd scripts to generate the plots shown in the manuscript. Links to the rendered HTML notebooks are below
• illustrations and output contain the figures

Code to generate figures

• Figure 1: Conceptual differences between variance-stabilizing transformations
• Figure 2: Benchmark results
• Figure 3: Computational expense
• Figure 4: Comparison of selected transformations

• plot_size_factor_effect_on_dim_reduction.Rmd generates Extended Data Fig. S1.
• plot_variance_stabilization_results.Rmd generates Extended Data Fig. S2.
• plot_bimodal_gene_histograms.Rmd generates Extended Data Fig. S3.
• plot_dataset_structure.Rmd generates Suppl. Figs. S1, Extended Data Fig. S5, and Suppl. Tabl. S1.
• plot_benchmark_results.Rmd generates Extended Data Figs. S6, S7, in addition to Fig. 2 and 3.
• plot_deep_sequenced_overlap.Rmd generates Extended Data Fig. S8.
• plot_clustering_results.Rmd generates Extended Data Fig. S9.
• plot_stratified_performance.Rmd generates Extended Data Figs. S10.