Gallery

library(ggvariant)

vcf_file <- system.file("extdata", "example.vcf", package = "ggvariant")
variants <- read_vcf(vcf_file)
#> ℹ Reading VCF: example.vcf
#> ✔ Reading VCF: example.vcf [26ms]
#> 
#> Loaded 19 variant records across 7 chromosomes.

This gallery collects one example of every ggvariant plot type against the bundled example VCF. See vignette("ggvariant") for a narrated walkthrough, and the function reference for the full argument list of each plot.

Lollipop plot

plot_lollipop(variants, gene = "TP53")

Consequence summary

plot_consequence_summary(variants, group_by = "gene", top_n = 6)

Mutational spectrum

plot_variant_spectrum(variants, facet_by_sample = TRUE)
#> Excluded 2 non-SNV records from spectrum plot.

Oncoprint / waterfall plot

plot_oncoprint() and plot_waterfall() are the same function under the two names used by different communities — “oncoprint” in the ComplexHeatmap/ cBioPortal vocabulary, “waterfall plot” in the GenVisR/maftools vocabulary. A waterfall plot and an oncoprint are the same visualisation: plot_waterfall() and plot_oncoprint() are aliases for one function and produce identical output — use whichever name your field uses.

The bundled example VCF has only two samples, too few to show the cascade staircase this plot is built around. The cohort below is illustrative synthetic data generated just for this article (not the package’s example data), sized like a real cohort so the staircase and the annotation track are both visible, with a clinical stage track added via annotation:

set.seed(42)

genes <- c("TP53", "KRAS", "PIK3CA", "EGFR", "BRCA1",
           "BRCA2", "APC", "PTEN", "ARID1A", "SMAD4")
gene_freq <- c(TP53 = 0.85, KRAS = 0.55, PIK3CA = 0.40, EGFR = 0.30,
               BRCA1 = 0.22, BRCA2 = 0.18, APC = 0.14, PTEN = 0.10,
               ARID1A = 0.07, SMAD4 = 0.05)
samples <- sprintf("COHORT_S%02d", 1:28)
consequences <- c("missense_variant", "stop_gained",
                   "frameshift_variant", "synonymous_variant")

cohort <- do.call(rbind, lapply(genes, function(g) {
  n_mut <- max(1, round(gene_freq[[g]] * length(samples)))
  data.frame(
    chrom = "chr1", pos = sample(1e6:2e6, n_mut),
    ref = "A", alt = "T", gene = g,
    sample = sample(samples, n_mut),
    consequence = sample(consequences, n_mut, replace = TRUE),
    stringsAsFactors = FALSE
  )
}))
cohort_variants <- coerce_variants(cohort)

clinical_stage <- data.frame(
  sample = samples,
  stage  = sample(c("I", "II", "III", "IV"), length(samples), replace = TRUE)
)

plot_waterfall(cohort_variants, top_n = 10, annotation = clinical_stage)

Tumour mutational burden

Reusing the same illustrative synthetic cohort from the oncoprint example above gives a more realistic view of per-sample burden across 28 samples than the two-sample example VCF would:

plot_tmb(cohort_variants)