Agent Skills
› aiming-lab/AutoResearchClaw
› scientific-visualization
scientific-visualization
GitHub提供基于matplotlib和seaborn的科研图表设计指南，涵盖期刊规格、无障碍配色、多面板布局及统计标注规范，确保生成符合出版要求的高质量科学可视化图像。
Trigger Scenarios
需要创建用于学术期刊投稿的科学图表
使用matplotlib或seaborn进行数据可视化并需符合出版标准
Install
npx skills add aiming-lab/AutoResearchClaw --skill scientific-visualization -g -y
SKILL.md
Frontmatter
{
"name": "scientific-visualization",
"metadata": {
"author": "researchclaw",
"version": "1.0",
"category": "writing",
"priority": "3",
"references": "adapted from K-Dense-AI\/claude-scientific-skills",
"trigger-keywords": "figure,plot,chart,visualization,matplotlib,seaborn,colorblind,publication",
"applicable-stages": "14,17,22"
},
"description": "Publication-ready scientific figure design with matplotlib and seaborn. Use when creating journal submission figures with proper formatting, accessibility, and statistical annotations."
}
Scientific Visualization Best Practice
Figure Design Principles
- Every figure must have a clear, self-contained message
- Minimize chartjunk: remove gridlines, background shading, and 3D effects
- Use direct labeling instead of legends when possible
- Remove top and right spines for cleaner appearance
- Ensure all text is readable at final print size (minimum 6pt font)
Journal Figure Sizing
- Single column: 3.3-3.5 inches (85-89 mm) wide
- 1.5 column: 4.5-5.5 inches (114-140 mm) wide
- Double column / full width: 6.5-7.1 inches (165-180 mm) wide
- Resolution: 300 DPI minimum for raster; prefer vector formats (PDF, EPS, SVG)
- Check target journal author guidelines for exact specifications
Colorblind-Safe Design
- Use colorblind-friendly palettes: seaborn "colorblind", Okabe-Ito, viridis, cividis
- NEVER rely on color alone — combine with shape, pattern, or line style
- Avoid red-green combinations; prefer blue-orange or blue-yellow contrasts
- Test figures with a colorblind simulator before submission
- Ensure figures work in grayscale for print journals
Multi-Panel Layouts
- Label panels with uppercase letters: (A), (B), (C) in bold, top-left corner
- Use consistent axis scales across panels when comparing related data
- Share axes where appropriate to reduce redundancy
- Maintain consistent font sizes and line widths across all panels
- Use
plt.subplots()withconstrained_layout=Truefor automatic spacing
Statistical Annotations on Figures
- Show individual data points alongside summary statistics (box + strip plots)
- Always include error bars; specify type in caption (SEM, SD, 95% CI)
- Use significance brackets with stars: * p<.05, ** p<.01, *** p<.001
- Annotate effect sizes or key statistics directly on the figure when helpful
- Never use bar charts for small-n data — use dot plots or box plots instead
Export and Quality Checklist
- Save in vector format (PDF/SVG) for line art; TIFF/PNG for photographs
- Embed fonts or convert text to outlines for cross-platform consistency
- Verify axis labels include units in parentheses: "Time (s)", "Force (N)"
- Ensure figure caption fully explains all symbols, abbreviations, and panels
- Check that color-coded elements match between figure and caption
Version History
- e2e23c9 Current 2026-07-25 07:48
