EasyPlotLib 0.3.0

A simple plotting library

EasyPlotLib

Matplotlib styles and helpers for publication-quality SCI-journal figures (Nature / Science / Cell / NEJM / Lancet / PNAS / AGU / AMS).

Install

pip install EasyPlotLib

Quick start

import EasyPlotLib as epl
import matplotlib.pyplot as plt
import numpy as np

# Style + journal column width + color palette in one call.
epl.journal_style("nat1", palette="nature", nrows=1, ncols=2)

fig, axs = plt.subplots(1, 2)
x = np.linspace(0, 2 * np.pi, 200)
for n, ax in enumerate(axs):
    for k in range(3):
        ax.plot(x, np.sin(x + k), label=f"series {k}")
    ax.set_xlabel("x")
    ax.set_ylabel("amplitude")
    ax.annotate(**epl.subplot_labels(n, "a"))      # bold panel labels a, b …

# Editable vector (PDF) + 600-dpi raster.
epl.save_pub(fig, "figures/fig1", formats=("pdf", "png"))

API

Function	Description
journal_style(key, palette=, base_style=, nrows=, ncols=, apply=)	Apply style + size + palette in one call. Returns the figsize dict.
figsizes(key, ...)	rcParams dict with a journal column figure size. Keys: nat1/2, aaas1/2, pnas1..3, agu1..4, ams1..4.
subplot_labels(n, style)	Args for ax.annotate(**...). Styles: a, A, (a), a), a.
geo_aspect(lon0, lon1, lat0, lat1)	Display aspect (w/h) of an equal-aspect lon/lat map: Δlon·cos(lat)/Δlat. Feed to row_layout / width_ratios.
row_layout(aspects, vmargin=)	For a 1-row map/plot mix: derive width_ratios (∝ aspect → equal heights) + a safe inverted_aspect_ratio so every panel fills its column.
clamp_colorbars(fig, (ax, cb), ...)	Clamp each ax=-attached colourbar's height to its (equal-aspect) panel instead of the taller cell.
set_palette(name, ax=None) / get_palette(name, n=)	Set the color cycle / return raw hex list.
PALETTES	Qualitative palettes: nature, science, nejm, lancet, jama, muted, semantic, nmi, comparison, imaging, bright.
SEMANTIC	Role-based colors (blue_main=hero, green_strong=gain, red_strong=drop, delta_up/down, neutrals, accents).
shades(color, n)	n shades of one hue, light→base — related methods (e.g. a baseline trio, stacked-area family).
alpha_ramp(color, n)	One hue at graduated opacity — the standard ablation encoding.
focal_palette(labels, focal, focal_color, other=)	Emphasise one focal series (saturated) while comparators recede (muted/grey/ordinal).
COLORMAPS	Recommended continuous colormaps for heatmaps/images: sequential, sequential_warm, diverging, grayscale.
save_pub(fig, path, formats=, dpi=)	Save to multiple formats with editable vector text.
cartopy_plot_tickmarks(ax, gl)	Clean lon/lat tick labels for cartopy maps.

Figure-style toolkit (correctness & legibility helpers)

Function	Description
set_frame(ax, style)	Per-axes spine visibility: open (bottom+left), boxed (all four), none.
bar_with_points(ax, x, ymat, labels, colors, ...)	Bar = mean; overlay raw points or an interval (errorbar='sd'/'ci95', the t-CI valid at small n).
strip_with_median(ax, groups, values, ...)	Jittered points + a bold median tick per group — show the distribution at small n.
goodness_arrow(ax, text=, loc=, axis=)	Upright "higher = better" direction-of-goodness cue in the margin.
end_of_line_labels(ax, xs, ys, labels, ...)	Direct-label each line at its right end instead of a legend box.
two_tier_label(name, meta)	Two-line "name\nmetadata" label string.
panel_crops(fig, dpi=)	{letter: (x0,y0,x1,y1)} pixel crop boxes per lettered panel in the saved png — drive the perceptual QA pass.

Multi-panel composition (tile panels on a mm column grid)

Function	Description
figure_outline_schema()	JSON schema for a panel_outline (12-col grid: claim, width_mm, ncol, row_heights_mm, panels[]).
grid_geom(outline) / panel_px(outline, letter) / panel_xy(outline, letter)	Resolve the outline to pixels; panel_px is the exact size to render each panel at.
compose_figure(outline, {letter: png}, out_path, letter_case=)	Tile per-panel PNGs onto the grid and stamp bold panel letters.
compose_crops(outline)	Per-panel crop boxes in the composed png (perceptual self-QA mirror of panel_crops).
apply_outline_revisions(outline, revisions)	Set of panel letters to regenerate after figure-level review revisions.

Color scheme guidance

Every recurring nature-style color pattern is abstracted into one small, shared API so a whole figure stays consistent:

• One restrained palette per figure. Use a unified family across panels rather than maximizing hue variety; reduce saturation before adding categories.
• Assign color by meaning, not by index. SEMANTIC reserves blue_main for the proposed/hero method and green/red for gains/drops. Never remap the same method to a different hue in another panel.
• Many-method comparison → get_palette("comparison"): soft pastel baselines with related methods sharing a hue family, and one saturated hero (put your method last).
• Related methods / one family → shades(color, n): graduated lightness of a single hue (baseline trios, stacked-area bands).
• Ablations → alpha_ramp(color, n): one hue, opacity carries the ordering.
• Continuous data (heatmaps, density, images) → a perceptual colormap from COLORMAPS — sequential for magnitude, diverging for signed values.
• Microscopy → imaging accents (cyan/magenta) on a black background.

Style notes

The bundled nature style uses 7 pt sans (TeX Gyre Heros / Helvetica), an open frame (no top/right spines), and export settings that keep text editable in Illustrator/Inkscape (pdf.fonttype=42, svg.fonttype=none) at 600 dpi.

Claude Code skill

A personal sci-figure skill (~/.claude/skills/sci-figure/) walks Claude through the publication workflow and is auto-invoked on plotting tasks in any project where EasyPlotLib is installed. This repo ships only the core helper library; the worked examples (chart gallery, cartopy/China maps, WRF nested-domain maps, layout grids) and their index live in that skill's examples/ folder, not here.