AI Can Write Code — Why Use FigureYa?

One-line answer: AI can write code, but AI doesn't guarantee the code is correct. Every FigureYa module is written by domain experts and validated by the community — it has authorship and peer review. What you get is not "code that looks like it works," but a verified analysis pipeline.

Three Problems with AI-Generated Code

1. Too Many Tools, No Way to Choose

Bioinformatics visualization involves hundreds of R packages: ggplot2, pheatmap, ComplexHeatmap, survminer, clusterProfiler, ggtree, circlize … each with countless parameters.

You ask an AI "draw a volcano plot" and it gives you code — but what package did it choose? Are the parameters correct? Is the statistical method appropriate? The AI won't tell you "I chose this package because this paper used it." It just picks something that looks reasonable.

Every FigureYa module clearly documents: which packages are used, why they were chosen, how parameters are set — because it was written by someone who has actually done omics analysis.

2. Results May Be Unreliable

AI-generated code has a hidden risk: the code runs, but the result is wrong.

• Wrong statistical test (using t-test when Wilcoxon is needed)
• Inappropriate normalization
• Default plot parameters that mislead (e.g., y-axis not starting at 0)
• Missing multiple-testing correction

These errors don't produce warnings — the code "works" but the conclusions are wrong. You might make downstream decisions based on incorrect results and waste time before discovering the error.

FigureYa modules have been tested by the community — not just "can it run," but "is the output correct."

3. No One Is Accountable for the Results

AI-generated code has no one to hold accountable for the results. When something goes wrong, the AI won't tell you where the error might be or whether the result is trustworthy.

FigureYa has:
- Authorship: Each module's .Rmd file records the author and source
- Community validation: 2,870+ users have used it; issues are reported and fixed via GitHub Issues
- Paper citation: FigureYa is published in iMetaMed (>doi:10.1002/imm3.70005), the methodology was peer-reviewed
- Version tracking: Modules have commit history; every change is traceable

FigureYa + AI = The Best Combination

FigureYa and AI are not substitutes — they complement each other:

Recommended Workflow

You describe your needs and data
      ↓
AI recommends from 300+ FigureYa modules
      ↓
You confirm the right template (author-backed, community-validated)
      ↓
AI adapts it to your data (paths, column mapping, type conversion)
      ↓
Figure generation built on a verified module — results you can trust

💡 The key distinction: don't let AI write code from scratch — let AI adapt on top of a verified FigureYa module. Reliability comes from FigureYa; flexibility comes from AI.

Where Does Trust Come From?

Authors: Domain Experts

FigureYa modules are not AI-generated templates — they are written by people who have actually done omics research:
- Each module corresponds to a real analysis scenario (e.g., pan-cancer immune infiltration, single-cell cell-type annotation, survival curve comparison)
- Parameter choices in the code are based on real data debugging experience, not default-value stacking
- Authorship is recorded in .Rmd files

Usage Scale: What Can and Cannot Be Counted

FigureYa's true usage is hard to pin down, but multiple sources paint the picture:

Traceable Data Sources

Academic Citations

The FigureYa paper (>doi:10.1002/imm3.70005) has been cited 9 times since publication. A Google Scholar search for "FigureYa" finds 40+ papers mentioning or acknowledging FigureYa:

• Some papers formally cite FigureYa in their methods section
• More papers acknowledge in their text: "We would like to thank FigureYa (Xiao Ya Hua Tu) for the figure technology support"
• Papers citing FigureYa have themselves been cited thousands of times (e.g., Deng et al. 2020 cited 142 times, Song et al. 2022 cited 161 times)

Untraceable Users

The actual number of users is far larger than what can be counted, because:

1. Download and use, no registration — cloning the repo, downloading a ZIP, or grabbing a single module from the gallery doesn't register anywhere
2. Many users don't cite — they used FigureYa code for figures but didn't mention it in their paper's methods or acknowledgments
3. Indirect use through platforms — some bioinformatics plotting platforms or tools integrate FigureYa code; end users don't know FigureYa is under the hood
4. Teaching spread — used as course material in training programs; students get the code directly
5. WeChat community — modules shared through WeChat groups, bypassing GitHub entirely

So if you've used FigureYa but didn't cite it — that's fine. You're part of the FigureYa community just like those who acknowledged it in their papers. Of course, mentioning FigureYa in your methods or acknowledgments helps more people discover the tool and keeps the maintainers motivated.

Paper: A Citable Academic Output

Lu, X. et al. (2025). FigureYa: A Standardized Visualization Framework for Enhancing Biomedical Data Interpretation and Research Efficiency. iMetaMed, 1, e70005. >https://doi.org/10.1002/imm3.70005

When you use a FigureYa module, you can cite this paper in your methods section — giving your visualization approach a traceable academic basis. AI-generated code cannot provide this kind of scholarly backing.

Next Steps

• 🎯 Not sure which module? How to Choose a FigureYa Module
• ▶️ How to run? How to Run FigureYa in Wisp Science
• 📖 Why Wisp Science? Why We Now Recommend Wisp Science
• 🌐 Community ecosystem: FigureYa Ecosystem