Publication-ready genomic variant plots in a few lines of R.
ggvariant fills a gap in the Bioconductor/CRAN ecosystem: a simple,
ggplot2-native package that takes you directly from a VCF file or data
frame to beautiful, customisable variant visualisations — without wrestling
with complex APIs or writing 50-line wrangling scripts.
| Need | Existing options | The problem |
|---|---|---|
| Lollipop plots | Gviz, karyoploteR | Very steep learning curve |
| Consequence summaries | maftools | Tightly coupled to MAF / cancer genomics |
| Mutational spectra | MutationalPatterns | Heavyweight; requires BSgenome |
| General VCF → ggplot | — | Nothing simple exists |
ggvariant gives both wet-lab biologists and experienced bioinformaticians
the same clean, ggplot2-idiomatic entry point.
# Install from GitHub (once released)
# install.packages("remotes")
remotes::install_github("josh45-source/ggvariant")library(ggvariant)
# 1. Load a VCF file
variants <- read_vcf("my_variants.vcf")
# 2. If you have a data frame instead (e.g. from Excel)
variants <- coerce_variants(my_df,
chrom = "Chr", pos = "Position",
ref = "Ref", alt = "Alt",
gene = "Gene", sample = "SampleID"
)plot_lollipop(variants, gene = "TP53")Add protein domain annotations:
tp53_domains <- data.frame(
name = c("Transactivation", "DNA-binding", "Tetramerization"),
start = c(1, 102, 323),
end = c(67, 292, 356)
)
plot_lollipop(variants, gene = "TP53", domains = tp53_domains)Colour by sample instead of consequence:
plot_lollipop(variants, gene = "TP53", color_by = "sample")# Stacked bar by sample
plot_consequence_summary(variants)
# Proportional
plot_consequence_summary(variants, position = "fill")
# Top 10 mutated genes
plot_consequence_summary(variants, group_by = "gene", top_n = 10)# 6-class SBS spectrum
plot_variant_spectrum(variants)
# Faceted by sample
plot_variant_spectrum(variants, facet_by_sample = TRUE)
# Raw counts, not proportions
plot_variant_spectrum(variants, normalize = FALSE)All plot functions accept interactive = TRUE to return a plotly object
for sharing with collaborators who don't use R:
plot_lollipop(variants, gene = "BRCA1", interactive = TRUE)Because every function returns a standard ggplot object, you can layer
on any ggplot2 or extension code:
library(ggplot2)
plot_lollipop(variants, gene = "KRAS") +
scale_colour_brewer(palette = "Set2") +
theme(legend.position = "bottom") +
labs(subtitle = "KRAS mutations in cohort X")Access palettes directly:
gv_palette("consequence") # named hex vector
gv_palette("spectrum") # COSMIC SBS colours- Minimal code — one function call per plot type
- Two entry points — VCF files and plain data frames
- ggplot2-native — every plot is a
ggplotobject; extend freely - Opinionated defaults — looks good out of the box; no mandatory config
- Progressive disclosure — simple API for novices, full control for experts
ggvariant/
├── R/
│ ├── ggvariant-package.R # Package documentation
│ ├── read_vcf.R # read_vcf() and coerce_variants()
│ ├── plot_lollipop.R # plot_lollipop()
│ ├── plot_functions.R # plot_consequence_summary(), plot_variant_spectrum()
│ └── utils.R # Theme, palettes, shared helpers
├── tests/
│ └── testthat/
│ └── test-core.R # Unit tests
├── inst/
│ └── extdata/
│ └── example.vcf # Bundled example VCF
├── DESCRIPTION
└── NAMESPACE
-
plot_oncoprint()— sample × gene mutation matrix -
plot_copy_number()— CNV segment visualisation -
plot_rainfall()— kataegis / mutation density along genome -
plot_tmb()— tumour mutation burden comparison across cohorts - BSgenome integration for automatic trinucleotide context extraction
- Shiny module for non-coding users
Pull requests are welcome. Please open an issue first to discuss proposed changes. All contributions should include tests.
MIT