Journal-ready figures

Journals reject figures for the same few reasons: wrong width, text too small or too large, missing fonts, low resolution. A SecretPlots preset encodes those rules, so a figure is the right size and fonts are embedded from the first draft.

import numpy as np
import pandas as pd
import SecretPlots as sp

sp.use(preset="nature")      # every plot from here on: 89 mm, 5–7 pt, 300 dpi

That one line is usually all you need. The rest of this page shows what a preset changes, how to choose widths, and a checklist for submission.

One figure, six journals

The same data and the same chain of options, with only .preset() changed. Notice how width, text size, line weights and aspect follow each destination's rules. (Previews are scaled to fit the page; the exported files have the exact physical size.)

rng = np.random.default_rng(7)
weeks = np.arange(0, 13)
cohort = pd.DataFrame(
    [
        {"week": w, "arm": arm, "patient": f"{arm}{i}",
         "hba1c": 8.4 - drop * (1 - np.exp(-w / 4)) + rng.normal(0, 0.22)}
        for arm, drop in (("Placebo", 0.2), ("Low dose", 0.8), ("High dose", 1.4))
        for i in range(20)
        for w in weeks
    ]
)
trial = (
    sp.line(cohort, x="week", y="hba1c", hue="arm", error="ci")
    .markers()
    .endlabels()
    .colors(["#999999", "#56B4E9", "#0072B2"])
    .labels(x="Weeks on treatment", y="HbA1c (%)")
)

NatureScienceCellIEEEPNASSpringer

trial.copy().preset("nature")

nature: 89 mm single column, 5–7 pt sans-serif, 300 dpi, PDF/EPS/SVG/TIFF.

trial.copy().preset("science")

science: 90 mm single column, 6–9 pt, lines at least 0.28 pt.

trial.copy().preset("cell")

cell: 85 mm single column, 6–8 pt Arial, lines at least 0.5 pt.

trial.copy().preset("ieee")

ieee: 3.5 in column, text sized to match a 9–10 pt paper body.

trial.copy().preset("pnas")

pnas: 87 mm single column, text at least 6 pt, lines at least 0.25 pt, 600 dpi.

trial.copy().preset("springer")

springer: 84 mm single column, 8–12 pt Helvetica/Arial.

Beyond journals, "slide-16x9", "poster", "a4", "letter", "screen" and "web" size the same figure for talks, posters, theses and the web.

Single or double column

Each journal preset names its allowed widths. Pass a name to .size(), or any explicit length such as "120mm" or "4.5in".

(
    trial.copy()
    .preset("nature")
    .size(width="double", height="60mm")
)

.size(width="double", height="60mm"): 183 mm, the full Nature page width.

Preset	Named widths	Text (pt)	Max height	dpi	Formats
nature	single 89 mm · double 183 mm	5–7	170 mm	300	pdf, eps, svg, tiff
science	single 90 mm · double 183 mm	6–9		300	pdf, eps, svg, tiff, png
cell	single 85 · one-and-half 114 · double 174 mm	6–8		300	pdf, eps, tiff, jpg
pnas	single 87 · one-and-half 114 · double 178 mm	≥ 6	225 mm	600	pdf, eps, tiff
aha	single 80 · one-and-half 127.5 · double 177.5 mm	6–10		600	tiff, eps, pdf
elsevier	minimal 30 · single 90 · one-and-half 140 · double 190 mm	≥ 6		500	eps, pdf, tiff, jpg
springer	single 84 · small 119 · double 174 mm	8–12	234 mm	600	eps, pdf, tiff
ieee	single 3.5 in · double 7.16 in	~9–10	8.8 in	600	pdf, eps, ps, png, tiff
acm	single 84.7 · double 177.8 mm	8–9		300	pdf, png
neurips	full 5.5 in	8–10		300	pdf, png

Each journal preset cites the publisher's author guidelines it was built from; see presets and themes for the links. Guidelines change, so check the current instructions before you submit.

What a preset enforces

Presets set limits as well as defaults. If you set a value outside the journal's range, SecretPlots keeps your choice and warns you:

>>> sp.box(df, x="dose", y="viability").theme(font_size=10).save("fig2.pdf")
LimitExceededWarning: font_size=10 breaks preset 'nature' limits (text 8.5-14.5 pt,
nature allows 5 pt to 7 pt); kept because you set it. Set
SecretPlots.config.enforce_limits = True to apply the limits instead.

Set sp.config.enforce_limits = True to clamp explicit values too, or sp.config.strict = True to turn every warning into an error (useful in CI for a paper's figure scripts). When a style looks unexpected, ask where it came from:

plot = trial.copy().preset("nature")
plot.explain("font_size")        # which layer set it: plot, preset or theme
plot.style_report()              # every resolved value and every adjustment

Colour that survives review

Around one in twelve men has a colour-vision deficiency, and many readers print in greyscale. Two habits cover both:

• Choose colour-blind-safe palettes. The default palette is one, and sp.colormaps() flags maps that are not (red-green, rainbow-like).
• Encode groups twice. .encode_groups() adds distinct dash patterns and markers, and the print-bw theme renders the whole figure in greys.

(
    sp.line(cohort, x="week", y="hba1c", hue="arm", error="ci")
    .theme("print-bw")
    .encode_groups()
    .endlabels()
    .labels(x="Weeks on treatment", y="HbA1c (%)")
)

.theme("print-bw") with .encode_groups(): each arm stays distinguishable without colour. (Explicit .colors() would override the theme's greys, so this chain leaves them out.)

Multi-panel figures

Most journals want a figure such as "Figure 2a–d" as one file. The most reliable workflow is to export each panel at its final physical size and assemble them in your layout tool (Illustrator, Inkscape, Affinity, or LaTeX subfigure) without rescaling:

half = {"width": "89mm", "height": "60mm"}     # two panels across a 183 mm page
panel_a = trial.copy().preset("nature").size(**half)
panel_b = trial.copy().preset("nature").size(**half).log("y")
panel_a.save("fig2a.pdf")
panel_b.save("fig2b.pdf")

Because each panel uses the same preset, text sizes and line weights match across panels. Add the bold panel letters (a, b, …) in the layout tool, where they are easy to align. To assemble the panels in matplotlib instead, draw each plot into your own axes with plot.render(ax=...): fonts, tick sizes and line weights follow the plot's preset, while the figure size and layout are yours, so create the figure at its final size; see data, formatting and export.

Submission checklist

Before you upload

• [x] Width matches a column width: .size(width="single" | "double").
• [x] Text is within the journal's range; no LimitExceededWarning was printed.
• [x] Format is vector (PDF/EPS/SVG) for plots; raster formats use the preset's dpi.
• [x] Fonts are embedded: SecretPlots saves PDF and EPS with TrueType (Type 42) fonts.
• [x] Error bars are defined in the legend: SD, SEM or 95% CI, and n.
• [x] Individual data points are shown for small samples (.points(), units=).
• [x] Statistics are reported: the test, the correction and the exact P values (.test_results()).
• [x] Gene and species names are italic (.ticklabels(italic=True)).
• [x] Colour is colour-blind safe and readable in greyscale.

Your own journal or lab style

A preset is a small TOML file. Copy the closest bundled one, adjust widths and limits, and register it:

sp.presets.register("my_journal.toml")
sp.use(preset="my_journal")

See presets and themes for the file format, and contributing if you'd like to add it to SecretPlots.