using CairoMakie
using DataFrames
using Statistics
using RandomBar Plot (Julia)
A bar plot compares quantities across discrete categories by their height. In biomedical research, bar plots are ubiquitous for reporting mean responses across treatment groups, cell-type composition per sample, and cohort summaries. Julia’s CairoMakie provides grouped (dodge) and stacked (stack) bar variants with publication-quality vector output.
Example
Setup
- System Requirements: Cross-platform (Linux/MacOS/Windows)
- Programming Language: Julia
- Dependencies:
CairoMakie,DataFrames,Statistics
Data Preparation
We simulate a drug-dose response experiment (four groups, replicate measurements) and a cell-type composition matrix (six samples).
Random.seed!(42)
groups = ["Vehicle", "Low dose", "Mid dose", "High dose"]
n_per_group = 40
raw = DataFrame(
Group = repeat(groups, inner = n_per_group),
Response = vcat(
2.0 .+ 0.9 .* randn(n_per_group),
3.1 .+ 0.9 .* randn(n_per_group),
4.6 .+ 0.9 .* randn(n_per_group),
6.2 .+ 0.9 .* randn(n_per_group),
),
)
agg = combine(groupby(raw, :Group),
:Response => mean => :Mean, :Response => std => :SD,
:Response => (x -> std(x) / sqrt(length(x))) => :SEM)
# Cell-type composition per sample (columns sum to 1)
comp = rand(4, 6)
comp ./= sum(comp, dims = 1)
cell_types = ["T cells", "B cells", "Macrophages", "NK cells"]
samples = ["S$i" for i in 1:6]6-element Vector{String}:
"S1"
"S2"
"S3"
"S4"
"S5"
"S6"Visualization
Basic Bar Plot with Error Bars
When bar heights summarize grouped measurements, always attach an error bar (SEM or 95% CI) with errorbars!.
fig = Figure(size = (480, 340))
ax = Axis(fig[1, 1], xlabel = "Treatment group", ylabel = "Mean response (a.u.)")
barplot!(ax, 1:4, agg.Mean; color = :slateblue, gap = 0.35)
errorbars!(ax, 1:4, agg.Mean, agg.SEM; linewidth = 1.2, whiskerwidth = 8)
ax.xticks = (1:4, groups)
hidespines!(ax, :t, :r)
fig
Grouped Bar Plot
A second factor is mapped to the dodge keyword, producing side-by-side bars within each category.
Random.seed!(43)
heights = vec([2.0, 2.3, 3.1, 3.3, 4.6, 4.8, 6.2, 6.4])
grp = repeat(1:4, inner = 2) # category id per bar
sex_ids = repeat(1:2, 4) # sub-group id per bar
sex_colors = [:plum, :steelblue]
fig = Figure(size = (540, 340))
ax = Axis(fig[1, 1], xlabel = "Treatment group", ylabel = "Mean response (a.u.)")
barplot!(ax, grp, heights; dodge = sex_ids, color = sex_colors[sex_ids], gap = 0.25)
ax.xticks = (1:4, groups)
labels = ["Female", "Male"]
elements = [PolyElement(polycolor = sex_colors[i]) for i in 1:2]
Legend(fig[1, 2], elements, labels, framevisible = false)
hidespines!(ax, :t, :r)
fig
Stacked Bar Plot
The stack keyword piles sub-group values on one bar — the standard display for per-sample composition.
n_types, n_samples = size(comp)
heights = vec(comp) # bar-major order
stack_ids = repeat(1:n_types, n_samples)
cat_ids = repeat(1:n_samples, inner = n_types)
type_colors = [:teal, :steelblue, :salmon, :olivedrab]
fig = Figure(size = (560, 340))
ax = Axis(fig[1, 1], xlabel = "Sample", ylabel = "Cell fraction")
barplot!(ax, cat_ids, heights; stack = stack_ids,
color = type_colors[stack_ids], gap = 0.15)
ax.xticks = (1:6, samples)
elements = [PolyElement(polycolor = type_colors[i]) for i in 1:n_types]
Legend(fig[1, 2], elements, cell_types, framevisible = false)
hidespines!(ax, :t, :r)
fig
References
- Makie development team. Makie.jl documentation — barplot. https://docs.makie.org/stable/reference/plots/barplot
- Danisch, S., & Krumbiegel, J. (2021). Makie.jl: Flexible high-performance data visualization for Julia. Journal of Open Source Software, 6(65), 3349.