Understanding Boxplots and Violin Plots

When to use boxplots and violin plots

Boxplots and violin plots are foundational exploratory views. They are useful when your main question is not yet “which formal model should I run?” but instead:

how are the cytokine values distributed?
do some groups look shifted, wider, or more variable than others?
are there obvious outliers or unusual shapes I should know before choosing a statistical test?

These are often the best first plots to inspect after Step 2 filtering.

When not to use boxplots and violin plots

These plots are less suitable when:

you want a compact many-variable summary rather than a full distribution view
you need a predictive or multivariate model
your next decision depends on effect size thresholds or model validation rather than raw spread

In those cases, error-bar plots, volcano plots, PCA, or supervised methods may answer the question more directly.

What the app is showing

Boxplots summarize each cytokine with quartiles, median, and potential outliers.

Violin Plots emphasize the full distribution shape and can optionally add Show Boxplot Overlay so you can see both the density shape and the boxplot summary together.

Both analyses can be ungrouped or split by selected categorical variables.

Which Step 4 arguments matter most

For both methods, the most important controls are:

Grouping Columns (Optional): whether the distributions are shown overall or split by group.
Bin Size: how many numeric variables are shown on one page.
Y-Axis Limits: whether to keep automatic scaling or force a shared scale across plots.

For violin plots specifically:

Show Boxplot Overlay: adds quartile and median structure inside the violin.

These decisions matter more than stylistic choices because they change what patterns are easy to see.

How to read the main outputs

Boxplots

Boxplots are best for quick summary reading:

the median shows the central value
the box shows the middle spread
the whiskers and isolated points help flag potential outliers

Use them when you want a cleaner, simpler overview.

Violin Plots

Violin plots are best when the shape itself matters:

wide regions indicate where values are more concentrated
narrow regions indicate fewer observations
asymmetry or multiple bulges can reveal skewness or multimodality

Use them when boxplots feel too compressed to describe the data shape.

Reading grouped views

When Grouping Columns (Optional) is used:

compare both center and spread
ask whether one group is simply shifted or whether the shape is genuinely different
look for strong overlap before over-interpreting apparent differences

Common cautions

Keep these limits in mind:

small groups can make violin shapes look more certain than they really are
auto-scaled y-axes can make panels look more different than they are
outliers may reflect biology, quality issues, or data-entry problems, so they deserve follow-up rather than automatic removal
grouped exploratory plots are not formal hypothesis tests

How to reproduce the result in the app

Filter the dataset to the groups and cytokines you want to inspect.
Choose Boxplots or Violin Plots.
Decide whether to use Grouping Columns (Optional).
Adjust Bin Size if the page is too dense or too sparse.
Leave Y-Axis Limits automatic unless you need consistent scales across pages.
For violin plots, turn on Show Boxplot Overlay if you want both shape and quartile summaries.