Strip plot
A strip plot draws every individual observation as a point along a categorical axis. Nothing is summarised, so the sample size and the exact shape of the distribution are both visible — which is exactly what a box or violin hides.
SELECT kuva_render(to_json({
'title': 'Expression by treatment',
'x_axis': {'name': 'group'},
'y_axis': {'name': 'expression'},
'series': [{
'type': 'strip',
'groups': groups,
'point_size': 2.5,
'style': {'jitter': 0.35},
'color': 'steelblue'
}]
})) AS chart
FROM (
SELECT list({'label': g, 'values': vals} ORDER BY g) AS groups
FROM (
SELECT "group" AS g, list(expression) AS vals
FROM read_csv_auto('{{DFK_BASE_URL}}data/samples.tsv')
GROUP BY "group"
)
);
Layout modes
style controls how points spread horizontally inside their slot:
style | Placement |
|---|---|
{"jitter": 0.3} | Random offsets, ±jitter of the slot width (the default) |
"swarm" | Non-overlapping — each point as close to the centre as it fits |
"center" | All points on the centre line, forming a vertical column |
jitter has a fixed seed, so the layout is reproducible; seed changes it.
A swarm traces the density of the distribution with its outline, which is the most readable of the three at
moderate sample sizes (roughly N < 200 per group):
SELECT kuva_render(to_json({
'title': 'Beeswarm',
'x_axis': {'name': 'group'},
'y_axis': {'name': 'expression'},
'series': [{
'type': 'strip',
'groups': groups,
'point_size': 3,
'style': 'swarm',
'color': 'steelblue'
}]
})) AS chart
FROM (
SELECT list({'label': g, 'values': vals} ORDER BY g) AS groups
FROM (
SELECT "group" AS g, list(expression) AS vals
FROM read_csv_auto('{{DFK_BASE_URL}}data/samples.tsv')
GROUP BY "group"
)
);
"center" stacks everything on one line, which turns the vertical packing itself into the density
estimate — gaps and clusters are unmistakable:
SELECT kuva_render(to_json({
'title': 'Centred stack',
'x_axis': {'name': 'group'},
'y_axis': {'name': 'expression'},
'series': [{
'type': 'strip',
'groups': groups,
'point_size': 2,
'style': 'center',
'color': 'steelblue'
}]
})) AS chart
FROM (
SELECT list({'label': g, 'values': vals} ORDER BY g) AS groups
FROM (
SELECT "group" AS g, list(expression) AS vals
FROM read_csv_auto('{{DFK_BASE_URL}}data/samples.tsv')
GROUP BY "group"
)
);
Horizontal orientation
horizontal swaps the axes: categories run down the y-axis and values along x (seaborn orient='h',
ggplot2 coord_flip). Jitter and swarm spread vertically within each row. It completes the
horizontal set alongside box and violin, so a box + strip pair can both be
flipped together.
SELECT kuva_render(to_json({
'title': 'Horizontal strip plot',
'x_axis': {'name': 'expression'},
'y_axis': {'name': 'group'},
'series': [{
'type': 'strip',
'groups': groups,
'style': 'swarm',
'point_size': 3,
'marker_opacity': 0.6,
'horizontal': true,
'color': 'steelblue'
}]
})) AS chart
FROM (
SELECT list({'label': g, 'values': vals} ORDER BY g) AS groups
FROM (
SELECT "group" AS g, list(expression) AS vals
FROM read_csv_auto('{{DFK_BASE_URL}}data/samples.tsv')
GROUP BY "group"
)
);
Combining with a box plot
A strip and a box series in the same series list share the axes, and the points are drawn on top of
the summary. Keep the points semi-transparent so the box stays legible underneath.
SELECT kuva_render(to_json({
'title': 'Box + strip',
'x_axis': {'name': 'group'},
'y_axis': {'name': 'expression'},
'series': [
{'type': 'box', 'groups': groups, 'color': 'steelblue', 'width': 0.7},
{'type': 'strip', 'groups': groups, 'point_size': 2,
'style': {'jitter': 0.25}, 'color': 'rgba(0,0,0,0.3)'}
]
})) AS chart
FROM (
SELECT list({'label': g, 'values': vals} ORDER BY g) AS groups
FROM (
SELECT "group" AS g, list(expression) AS vals
FROM read_csv_auto('{{DFK_BASE_URL}}data/samples.tsv')
GROUP BY "group"
)
);
The box gives you Q1 / median / Q3; the points are what reveal that Drug_B is really two sub-populations
the box has averaged into one.
Per-group colours
colors colours each group by position, falling back to the uniform color past the end of the list.
SELECT kuva_render(to_json({
'title': 'Per-group colours',
'x_axis': {'name': 'group'},
'y_axis': {'name': 'expression'},
'series': [{
'type': 'strip',
'groups': groups,
'colors': ['steelblue', 'tomato', 'seagreen', 'goldenrod', 'mediumpurple'],
'point_size': 3,
'style': {'jitter': 0.3}
}]
})) AS chart
FROM (
SELECT list({'label': g, 'values': vals} ORDER BY g) AS groups
FROM (
SELECT "group" AS g, list(expression) AS vals
FROM read_csv_auto('{{DFK_BASE_URL}}data/samples.tsv')
GROUP BY "group"
)
);
Per-point colours
A group also takes point_colors, one colour per point, when an observation belongs to a category and
that is what you want to show. The legend is not updated automatically — use one series per category if you
need labelled legend entries.
WITH d AS (
SELECT "group" AS g, expression
FROM read_csv_auto('{{DFK_BASE_URL}}data/samples.tsv')
WHERE "group" = 'Control'
)
SELECT kuva_render(to_json({
'title': 'Per-point colours',
'x_axis': {'name': 'sample'},
'y_axis': {'name': 'expression'},
'series': [{
'type': 'strip',
'groups': [
{'label': 'Control',
'values': (SELECT list(expression ORDER BY expression) FROM d),
'point_colors': (SELECT list(CASE WHEN expression > 5 THEN '#c44e52' ELSE '#4c72b0' END
ORDER BY expression) FROM d)}
],
'point_size': 4,
'style': 'swarm'
}]
})) AS chart;
Points past the end of point_colors fall back to the group or uniform colour. point_shapes does the
same for marker shapes.
Marker opacity and stroke
At high density the default solid fill merges into a bar. marker_opacity makes denser bands darker and
marker_stroke_width outlines each point in its fill colour, so individual observations stay countable.
SELECT kuva_render(to_json({
'title': 'Semi-transparent markers',
'x_axis': {'name': 'group'},
'y_axis': {'name': 'expression'},
'series': [{
'type': 'strip',
'groups': groups,
'point_size': 4,
'style': {'jitter': 0.3},
'color': 'steelblue',
'marker_opacity': 0.25,
'marker_stroke_width': 0.7
}]
})) AS chart
FROM (
SELECT list({'label': g, 'values': vals} ORDER BY g) AS groups
FROM (
SELECT "group" AS g, list(expression) AS vals
FROM read_csv_auto('{{DFK_BASE_URL}}data/samples.tsv')
GROUP BY "group"
)
);
Fields
| Field | Type | What it sets |
|---|---|---|
groups | group[] | Required. One column per group, each {label, values, point_colors?, point_shapes?}. |
colors | string[] | Per-group colours, matched to groups by position. |
point_size | number | Point radius in pixels (default 4). |
style | string | object | {"jitter": 0.3} · "swarm" · "center". |
seed | integer | RNG seed for the jitter positions (default 42) — keeps output reproducible. |
marker_opacity | number | Fill alpha: 0 is hollow, 1 solid. |
marker_stroke_width | number | Outline width, drawn in the fill colour. |
horizontal | boolean | Categories on the y-axis, values on x; jitter/swarm spread vertically. |
color and legend come from series & shared fields.
Notes
- Every group needs at least one value; an empty
groupslist is an error. colorsis matched by position, so its order must matchgroups.point_colors/point_shapesare per group and per point; shorter lists fall back to the group colour and shape."center"is the densest mode and the one that scales best to very large samples.
See also
- kuva — Strip plot — the plotting library's own reference for this chart.
- Box plot — the summary you would overlay it on.
- Violin plot — the density shape.
- Raincloud — all three at once.
- Legends → Hand-written entries — a key for colours that live in the data rather than in a series.