dysonsphere 3.3.0

An Altair theme and chart utility library for publication-ready scientific figures.

dysonsphere

dysonsphere is an altair utility library for publication-ready scientific figures in Python, offering:

• An attractive, cohesive, and sensible default configuration with a single invocation of ds.theme().
• Perceptually uniform palettes, including those from popular data visualization libraries.
• Shareable styles through a simple dysonsphere.toml config file - tune the theme once and reuse it across projects.

dysonsphere also comes with several intuitive and powerful utilities for altair charts; some highlights include:

• The ability to quickly and easily layer altair charts with annotations like reference lines, shades, text, and data labels.
• Statistical inference with scipy, layered directly onto alt.Chart and conveniently exported as metadata in your saved chart:
  • Omnibus tests with effect sizes.
  • Brackets and p-values for pairwise and post hoc comparisons.
  • Correlations with fit lines.
• Multilabels that allow for rich annotations of categorical labels, e.g. multi-condition axes and sample sizes.
• Self-documenting exports for reproducible figures:
  • ds.save() writes a corrected SVG, print-ready PNG, interactive HTML, and/or Vega-Lite JSON, while also embedding the output file with its provenance: environment versions and sha256 checksums fingerprint both the Vega-Lite spec and the underlying data.
  • ds.read() recovers the statistics report, metadata, and even the original data from a saved figure.
  • ds.load() rebuilds the chart from its JSON.

dysonsphere is extensible: you can develop and install packages of field-specific plotting toolkits (e.g. molecular biology, astronomy) for altair, which plug into the same theme, palettes, and export pipeline that core dysonsphere provides.

Installation

# with pip
pip install dysonsphere

# with uv
uv pip install dysonsphere

# add as a project dependency
uv add dysonsphere

Requires Python 3.11+. Every function that takes a DataFrame accepts polars or pandas.

Dependencies:

• altair>=5.5.0
• numpy>=1.26.0
• polars[pyarrow]>=1.19.0
• scipy>=1.11.0
• vl-convert-python>=1.7.0

Quick start

import altair as alt
import polars as pl
import dysonsphere as ds

ds.theme()  # apply the dysonsphere theme to all charts in the session

df = pl.DataFrame({"x": [1.2, 2.4, 3.1, 4.8], "y": [0.9, 2.2, 2.8, 4.4]})

chart = (
    alt.Chart(df)
    .mark_point()
    .encode(
        x="x:Q",
        y="y:Q",
        color=alt.Color("y:Q", scale=alt.Scale(range=ds.palette("blues"))),
    )
)

ds.save(chart, "myplot") # writes myplot.svg + myplot.json

Documentation

Documentation, examples, palettes, and an interactive chart studio at dkkung.github.io/dysonsphere