The R package sigvar implements signature variability analysis, a framework for the analysis of mutational signature activities within and across cancer samples. This R package accompanies the paper “Variability of mutational signatures is a footprint of carcinogens’’ by Morrison et al.; please refer to the paper for more details on the methods presented in this package.

The sigvar package contains two core functions to perform signature variability analysis:

• sigvar: Compute the within-sample diversity and across-sample heterogeneity of mutational signature activity in one or multiple populations of samples

• sigboot: Use bootstrapping to statistically compare the within-sample diversity and across-sample heterogeneity of the mutational signature activity between two or more groups of samples

sigvar also includes accessory functions for the visualization of mutational signature data, such as:

• plot_SBS_spectrum: Plot the SBS mutational spectrum of one or more samples of mutational signatures

• plot_signature_prop: Plot the relative activities of mutational signatures in each sample as a stacked bar plot

• plot_dots: Plot the mean mutational signature contributions of one or more groups of samples

You can install the development version of sigvar from GitHub with:

# Ensure devtools is installed
if (!requireNamespace("devtools", quietly = TRUE)) install.packages("devtools")

# Install the package from GitHub
devtools::install_github("MaikeMorrison/sigvar", dependencies = TRUE, build_vignettes = TRUE)

The package will ultimately be available on Bioconductor. When it is, you will be able to install it with:

# Ensure BiocManager is installed
if (!requireNamespace("BiocManager", quietly = TRUE)) install.packages("BiocManager")

# Install the package from Bioconductor 
BiocManager::install("sigvar")

Installation time ranges from 1 to 5 minutes depending on whether dependencies also need to be installed. Run time is expected to be a few minutes on a typical desktop computer.

The package has been tested on R version 4.1.2 on a Redhat Linux platform and a Windows 10 Pro platform. The package is available under the MIT license.

A tutorial on the usage of sigvar is available in the tutorial vignette, which is available at this link or via the following R code after package installation:

vignette("tutorial", package = "sigvar")

The run time of the tutorial is under 5 minutes.

There are also several vignettes that reproduce figures and analyses from Morrison et al. These vignettes can be accessed via the following R code after package installation:

vignette("carcinogen_figure", package = "sigvar")
vignette("Drivers", package = "sigvar")
vignette("geography_incidence_figure", package = "sigvar")
vignette("supplement_carcinogen_figure_altered", package = "sigvar")

dplyr, readr, ggplot2, rlang, tidyr, stringr, ggh4x, glue, ggtext, ggforce, scales, GenomicFeatures, GenomeInfoDb, BSgenome.Hsapiens.UCSC.hg38, BSgenome.Hsapiens.UCSC.hg19, BSgenome.Mmusculus.UCSC.mm10, Biostrings, rtracklayer, TxDb.Hsapiens.UCSC.hg38.knownGene, TxDb.Mmusculus.UCSC.mm10.knownGene, lifecycle