Lab Miniguide - How to Choose Biology Figures Like an Artist
So, it’s 8 pm. Your next presentation is tomorrow. And you’ve got a pile of GraphPad Prism data staring at you to turn into figures.
You’re thinking about throwing in bar charts everywhere—but you know this won’t show the depth of your project.
In this mini-guide, Wildtype One compiled common mistakes and best practices to turn your data into clear, beautiful, publication-ready figures that will impress your team and reviewers.
Main points to cover:
Why not a bar chart
Figure-data match
How to have a clear design
How to make the figure readable for everyone
Good vs. bad practices (From the trenches)
Align with journal standards
TLDR summary
🧫 Join our network of 400+ elite researchers by signing up for Wildtype One’s FREE newsletter.
1. Why not use a bar chart for all figures?
Because a bar chart is a summary, not a story.
In biological data, the story is in the nuance.
A bar graph with mean ± SEM masks data distribution. If your n < 10, the chart would be misleading (more about this in the next section).
2. Figure-data match
To express itself fully, each experiment needs a different chart.
Here are four common examples of different types of data and what figure to match with each:
Continuous data (small n)
Use dot plots or scatter plots with individual data points, overlaid with mean or median lines. Prism makes these easy. R's ggplot2 handles it beautifully as well.
Example: Measuring tumor volume in mice after treatment.
Continuous data (large n)
Too many points can now clutter dot plots. To summarize shape and spread, use box plots or violin plots. Showing frequency? Use histograms.
Example: Single-cell RNA-seq of 500 cells to quantify expression of a certain gene.
Paired data (before/after)
Spaghetti plots work best here (paired dot plots with connecting lines) show directionality.
Example: Measuring patient blood biomarker levels before and after treatment.
Categorical data
Bar graphs for proportions, stacked bars for subcategories. Avoid pie charts unless your data is a single, simple proportion.
Example: Flow cytometry classification (sorting) of immune cell types (e.g., % of CD8+, CD4+, Treg, NK cells).
Time series or dose response
Line graphs. Add error bars or shaded confidence intervals.
Example: Measuring phosphorylation levels of ERK (p-ERK) over time after EGF stimulation.
Also, keep in mind:
1. Stats indicators - Use clear asterisks (p<0.05, *p<0.01) or labeled comparisons. Clarify in legends. Avoid cramming too many p-values.
2. Show variability thoughtfully:
Use SD or IQR over SEM unless showing precision.
Clarify in the legends what your error bars represent.
3. How to have a clear design
A figure needs to answer one single question. Ask: What message is behind this figure? Answer with a single message. Build it around that.
Remove chartjunk, like gridlines, 3D effects, and shadows.
Remove neon gradients and Comic Sans fonts.
Axes need units. Legends need clarity. Use descriptive group names (e.g., "WT" vs. "KO" is fine, but better if spelled out).
Use the same fonts across figures.
If you have a panel with different charts, make sure they have the same scale (Graphpad Prism sets the axes automatically, if two similar charts have different scales, switch to manual and adjust them).
Start axes at zero (especially bar graphs). Keep aspect ratios honest.
Stick to 4–6 colors, but don’t use red/green for opposite samples.
4. How to make the figure readable for everyone
One in 12 men is color-blind.
So if you're presenting at a conference with 500 male researchers, 42 will NOT be interested in your low-contrast figures!
To make your figures readable to everyone:
Avoid red-green-only schemes.
Use contrasting colors (e.g., orange/blue or purple/green.)
If you’re ever in doubt, pick colors that are opposite to each other on the color wheel
Add shape or pattern redundancy
Print it or convert it to black and white—if your figure is unreadable in grayscale, rethink your colors
For fonts and labels:
Use a minimum 8pt font (final print size)
Dark lines, legible fonts
Avoid white-on-light-gray labels
5. Good vs. Bad practices (from the trenches)
Let’s lighten up and look at some do’s and don’ts—inspired by real anecdotes and complaints from the scientific community:
6. Align with journal standards
File types and resolution:
Use vector graphics (PDF, SVG, EPS) for plots.
Images at 300 dpi (TIFF for microscopy).
Figure layout:
Use A, B, C panels logically.
Don’t cram too many subplots.
Font sizes:
8–12 pt when shrunk to journal column width.
Show raw data when possible:
Journals like Nature, Cancer Cell, JCO now expect raw data points.
Graphical abstracts:
Simpler, schematic-friendly version of your paper
Keep it minimal and message-focused
7. TLDR Summary
