CellBioStats is a web application that simplifies the creation of publication-quality SuperPlots and performs statistical analysis on hierarchical data from common experimental results in cell biology, molecular biology, and biochemistry, helping you visualize complex datasets and avoid pitfalls like pseudoreplication.

The current version is still limited in terms of case uses, but the idea is the project to grow to be more comprehensive.

Features

• Interactive SuperPlots: Generates publication-quality SuperPlots, showing individual data points, replicate means, and overall treatment means with standard error (SEM).
• Automated Statistical Analysis: Automatically performs normality checks, variance homogeneity tests, and selects the appropriate statistical test for your data.
• Hierarchical Statistics: Avoids pseudoreplication by correctly performing statistical tests on replicate means, not on raw technical measurements.
• Paired & Unpaired Data: Handles both independent (unpaired) and repeated measures (paired) experimental designs.
• Data Upload: Supports both .csv and .xlsx file formats.
• Plot Customization: Allows customization of plot aesthetics (font size, color schemes, marker size, and plot templates).
• Downsampling for plot visualization: Option to display a random subset of your data points to keep plots clean and responsive with very large datasets.
• Export Results: Download the detailed statistical summary as a .txt file and the plot as a .png, .jpeg, .svg or .pdf file directly from the app.

Getting Started

Live App

CellBioStats is deployed on Render. Visit the live app

Run Locally

1. Clone the repository:

   git clone https://github.com/brunicardoso/CellBioStats.git
   cd CellBioStats

2. Create a virtual environment and install dependencies:

   python -m venv venv
   source venv/bin/activate   # On Windows: venv\Scripts\activate
   pip install -r requirements.txt

3. Run the application:

   uvicorn backend.main:app --reload

4. Open your web browser and navigate to http://localhost:8000.

How to Use the App

1. Upload Data: Drag and drop (or click to browse) your .csv or .xlsx file. You can try it on our sample data
2. Map Columns: Select the appropriate columns from your file for Treatment, Value, and Replicate.
3. Select Test Type: (Optional) If your experiment uses a repeated measures or paired design, toggle "Use paired/repeated measures tests". Read this if you are not sure about choosing paired or unpaired tests
4. Analyze: Click the "Analyze" button.
5. Customize: Expand "Plot customization" to adjust appearance (font size, colors, marker size, template, downsampling). Click "Analyze" again to update.
6. Download: Use the download buttons to save the plot (PNG/JPEG/SVG/PDF) or statistical summary (.txt).

Input Data Format

The application expects your data to be in a specific format, like in the example below. You must have at least three columns:

• Treatment Column: Identifies the different experimental groups (e.g., 'Control', 'Drug A').
• Value Column: Contains the numeric measurement data (the dependent variable; e.g., cell size, expression level, etc).
• Replicate Column: Identifies the independent experimental replicates (e.g., experiment number of independent replications done in different days or animal ID).

Here is an example of a valid data structure:

Treatment	Cell_size	Replicate
Control	10.5	1
Control	11.2	1
Control	12.1	2
Control	11.8	2
Drug A	15.3	1
Drug A	14.9	1
Drug A	16.5	2
Drug A	17.1	2

Statistical Methodology

CellBioStats is designed to perform statistically sound analysis by respecting the hierarchical nature of typical biological data.

1. Data Aggregation: All primary statistical tests are performed on the means of each replicate, not on the raw technical measurements. This avoids pseudoreplication and ensures that the statistical power reflects the number of independent experiments.
2. Assumption Checks:
   • Normality: The Shapiro-Wilk test is run on the replicate means for each treatment group.
   • Homoscedasticity (Equal Variances): Levene's test is run to check for equality of variances across groups.
3. Automated Test Selection: Based on the number of groups, the experimental design (paired/unpaired), and the results of the assumption checks, the app automatically selects the most appropriate statistical test.

# of Groups	Design	Assumptions Met (Normal & Homoscedastic)	Assumptions Not Met
2	Unpaired	Student's t-test	Mann-Whitney U test
>2	Unpaired	One-way ANOVA + Tukey HSD post-hoc	Kruskal-Wallis + Mann-Whitney U post-hoc (Bonferroni)
2	Paired	Paired t-test	Wilcoxon signed-rank test
>2	Paired	Repeated Measures ANOVA + Paired t-test (Bonferroni)	Friedman test + Wilcoxon post-hoc (Bonferroni)

Architecture

• Backend: FastAPI serving REST API endpoints and static files
• Frontend: HTML + Tailwind CSS + vanilla JavaScript + Plotly.js
• Deployment: Render via render.yaml

Built With

• FastAPI & Uvicorn
• Plotly / Plotly.js
• Tailwind CSS
• Pandas
• SciPy & Statsmodels

References

1. Lord, S. J., Velle, K. B., Mullins, R. D., & Fritz-Laylin, L. K. (2020). SuperPlots: Communicating reproducibility and variability in cell biology. Journal of Cell Biology, 219(6). https://doi.org/10.1083/jcb.202001064

2. Pollard, D. A., Pollard, T. D., & Pollard, K. S. (2019). Empowering statistical methods for cellular and molecular biologists. Molecular Biology of the Cell, 30(12), 1359-1368. https://doi.org/10.1091/mbc.e15-02-0076

Cite our work

Bruni-Cardoso, A. (2025). CellBioStats, an application for robust visualization and statistical analysis of data from cell and molecular biology experiments

License

This project is licensed under the MIT License - see the LICENSE file for details.