Stylia: decent scientific plot styles

Stylia provides predefined Matplotlib styles, color palettes, and figure utilities for producing publication-quality scientific figures. Designed for the Ersilia Open Source Initiative, but works for any scientific Python project.

Article style (default — NPG palette, black structural elements)

Ersilia style (Ersilia brand palette, plum structural elements)

---

Installation

pip install stylia

Importing stylia automatically applies global Matplotlib style settings:

import stylia

---

Table of Contents

• Format and style
• Named colors
• Continuous colormaps
• Categorical palettes
• Figures
• Sizes and constants
• Tutorial notebook

---

Format and style

stylia.set_format("print")   # default — compact, 7.09 in wide
stylia.set_format("slide")   # larger fonts and markers, 13 in wide

stylia.set_style("article")  # default — structural elements in black
stylia.set_style("ersilia")  # structural elements in Ersilia plum

Both update matplotlib.rcParams globally and can be changed at any point.

---

Named colors

ArticleColors

Modern palette spanning the full hue wheel for maximum distinctness.

	Name	Hex
	crimson	#E63946
	tangerine	#F4845F
	amber	#FCBF49
	lime	#6BBF59
	turquoise	#2EC4B6
	cobalt	#457B9D
	periwinkle	#6C5CE7
	orchid	#B05CC8
	fuchsia	#E91E8C
	silver	#A0A0A0

from stylia import ArticleColors

nc = ArticleColors()
ax.scatter(x, y, color=nc.crimson)
ax.scatter(x, y, color=nc.get("cobalt", alpha=0.4))
ax.scatter(x, y, color=nc.get("turquoise", lighten=0.3))

ErsiliaColors

Official Ersilia brand palette.

	Name	Hex
	plum	#50285A
	purple	#AA96FA
	mint	#BEE6B4
	blue	#8CC8FA
	yellow	#FAD782
	pink	#DCA0DC
	orange	#FAA08C
	gray	#D2D2D0

NamedColors (style-aware)

NamedColors resolves to ArticleColors or ErsiliaColors based on the active style:

nc = stylia.NamedColors()   # ArticleColors or ErsiliaColors depending on set_style()

---

Continuous colormaps

Four families, each with presets for the article (NPG) and ersilia palettes. All share fit(data) / transform(data, alpha=, lighten=) / sample(n).

FadingColormap

Near-white → single hue. Good for density or strictly positive data.

Article presets (NPG colors)

	Preset	Range
	crimson	blush → vivid red (default)
	cobalt	pale sky → steel blue
	turquoise	pale mint → teal-cyan
	orchid	pale lavender → orchid
	lime	pale green → lime

Ersilia preset

	Preset	Range
	plum	pale lilac → deep plum

SpectralColormap

Multi-hue warm → cool. Good for ordered data where the full range matters.

	Preset	Range
	npg (article)	crimson → amber → turquoise → periwinkle → fuchsia
	ersilia	orange → yellow → mint → blue → purple

DivergingColormap

Two hues through a light center. Good for data diverging around a meaningful midpoint.

Article presets (NPG colors)

	Preset	Range
	crimson_cobalt	red ↔ steel blue
	amber_periwinkle	amber ↔ blue-violet

Ersilia presets

	Preset	Range
	plum_mint	deep plum ↔ mint
	purple_orange	lavender ↔ peach

CyclicColormap

Wraps back to its starting color. Good for phase or angle data.

	Preset	Cycle
	npg (article)	crimson → tangerine → lime → turquoise → orchid → crimson
	ersilia	orange → yellow → mint → blue → purple → pink → orange

Usage

from stylia import FadingColormap, DivergingColormap

ccm = FadingColormap("turquoise")
ccm.fit(data)
colors = ccm.transform(data)                 # list of RGBA tuples
colors = ccm.transform(data, alpha=0.6)     # with alpha
colors = ccm.transform(data, lighten=0.3)   # lightened
swatches = ccm.sample(8)                    # 8 evenly-spaced swatches

dcm = DivergingColormap("crimson_cobalt", ascending=False)
dcm.fit(data)
colors = dcm.transform(data)

---

Categorical palettes

Two built-in palettes match the default styles. Three additional bonus palettes are available for specific needs (colorblind accessibility, soft aesthetics).

from stylia import CategoricalPalette

pal = CategoricalPalette()       # "npg" (default) or "ersilia"

colors = pal.get(5)              # 5 maximally distinguishable colors
colors = pal.get(20)             # >palette size: interpolated as a colormap
color  = pal.next()              # draw one at a time (advances internal counter)

Default palettes

npg — redesigned for maximum hue coverage (article style default)

ersilia — Ersilia brand (ersilia style default)

Bonus palettes

okabe — Okabe–Ito (colorblind-safe)

tol — Paul Tol Bright (colorblind-safe)

pastel — soft pastels

---

Figures

import stylia

stylia.set_format("print")
stylia.set_style("article")

fig, axs = stylia.create_figure(2, 1, width=0.8, height=0.3)

ax = axs.next()
ax.scatter(x, y, color=nc.crimson, s=stylia.get_markersize())

ax = axs.next()
ax.bar(groups, values, color=pal.get(len(groups)))

stylia.save_figure("figure.pdf")

create_figure accepts width and height as fractions of the format's base size. axs.next() steps through panels in order.

---

Sizes and constants

All constants resolve automatically to the correct value for the active format — just read stylia.SIZE, stylia.FONTSIZE, etc. after calling set_format(). The explicit SLIDE_* variants are also available if you need to reference both values at once.

width and height in create_figure are fractions of SIZE.

Constant	print	slide	Use
SIZE / SLIDE_SIZE	7.09 in	13 in	figure width basis
FONTSIZE_SMALL / SLIDE_FONTSIZE_SMALL	5 pt	8 pt	tick labels, annotations
FONTSIZE / SLIDE_FONTSIZE	6 pt	10 pt	axis labels, legend
FONTSIZE_BIG / SLIDE_FONTSIZE_BIG	8 pt	13 pt	panel titles
MARKERSIZE_SMALL / SLIDE_MARKERSIZE_SMALL	5	8	dense scatter (s=)
MARKERSIZE / SLIDE_MARKERSIZE	10	15	standard scatter (s=)
MARKERSIZE_BIG / SLIDE_MARKERSIZE_BIG	30	45	highlighted points (s=)
LINEWIDTH / SLIDE_LINEWIDTH	0.5	0.75	lines, spines
LINEWIDTH_THICK / SLIDE_LINEWIDTH_THICK	1	1.5	emphasis lines

Use stylia.get_markersize() to get the format-aware marker size at runtime ("small", "normal", or "big").

---

Tutorial notebook

An end-to-end walkthrough of all stylia features is available in notebooks/stylia_tutorial.ipynb. It covers format and style setup, named colors, all four colormap families, categorical palettes, and figure creation.

---

Disclaimer

Stylia is designed for internal use across Ersilia projects and is shared openly in case it is useful to others. It is not a general-purpose plotting library — for that, see Matplotlib or seaborn.

---

About Us

Stylia is developed and maintained by the Ersilia Open Source Initiative, a non-profit organisation dedicated to providing open-source AI/ML tools for infectious disease research in the Global South.

Visit us · GitHub