nature-figure

Create, audit, and export submission-grade scientific figures in Python or R for Nature-family journals.

Updated Jul 5, 2026
One-click install
npx skills add https://github.com/huaibovip/research-marketplace --skill nature-figure-huaibovip
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: nature-figure
Source: https://github.com/huaibovip/research-marketplace/tree/main/plugins/nature-skills/skills/nature-figure
Command: npx skills add https://github.com/huaibovip/research-marketplace --skill nature-figure-huaibovip

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve? Producing journal-ready scientific figures requires juggling plotting code, journal-specific formatting rules, data integrity, and pre-submission QA; this Skill turns raw data, legends, or manuscript claims into compliant multi-panel figures with reproducible scripts and audit records. ## Core Features & Use Cases - Publication-grade plotting in Python or R: Generates matplotlib/seaborn or ggplot2/patchwork/ComplexHeatmap scripts with a persisted backend preference, editable SVG text, and SVG/PDF/TIFF export at 300 dpi. - Journal contract enforcement: Applies flagship Nature and Nature Machine Intelligence rules for main displays, Extended Data, legend word limits, resolution, and source-data traceability. - Automated QA pipeline: Runs static validation on plotting source, audits exported PDFs for sub-5pt glyph sizes, and enforces panel-by-panel visual review before delivery. - AI schematic route: Optionally drafts graphical abstracts and mechanism diagrams via the OpenRouter GPT Image 2 API with policy gating, disclosure, and provenance tracking. - Use Case: A researcher provides a CSV of experimental results and asks for a Nature-style multi-panel figure; the Skill writes the Python script, exports editable SVG/PDF, maps source data, and delivers a pre-submission QA record. ## Quick Start Ask the assistant to turn your dataset into a Nature-style multi-panel figure, stating whether you prefer Python or R and the target journal.

Frequently Asked Questions about nature-figure

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I make a Nature-style multi-panel figure in Python?

Provide your data, target journal, and output format, then choose Python as the backend. The Skill writes a matplotlib script using publication rcParams, editable SVG text, and panel labels, then exports SVG/PDF/TIFF and runs QA checks before delivery.

Should I use Python or R for scientific plotting?

Choose R for ggplot2, ComplexHeatmap, Seurat, or survival-analysis workflows, and Python for custom matplotlib layouts, image plates, or NumPy/Pandas pipelines. The Skill saves your first choice as the default and reuses it for later figures.

What are the Nature figure requirements for submission?

Flagship Nature separates initial-review files from final main and Extended Data production contracts, with a legend limit under 250 words. Nature Machine Intelligence allows six main displays, up to ten Extended Data items, 300 dpi, and 180 mm width with source data.

Can AI generate a graphical abstract for my paper?

Yes, through the explicit OpenRouter GPT Image 2 route, which produces a concept draft rather than a data figure. You must verify the target journal's AI policy, run human scientific review, and keep disclosure and provenance records before submission.

Why does my exported PDF have fonts below the journal minimum?

Mathtext superscripts and subscripts can shrink below 5 pt even when the base fontsize is compliant. Run the included audit_pdf_text.py script on the exported PDF to detect undersized Tf glyph runs and adjust labels accordingly.

When should I not use this figure-making skill?

Avoid it for interactive dashboards, statistics-only analysis, data cleaning, literature review, pure photo editing, or Illustrator/Figma-first infographics without manuscript-figure intent. It is scoped to data-driven or schematic figures destined for journal submission.