Designing Scientific Figures

         

The old adage says “a picture is worth a thousand words.” If that is the case, then a picture should receive the same amount of care and attention as those thousand words. Yet, scientific figures are often secondary concerns, verbatim regurgitations of graphs generated by data processing software programs without any additional design considerations. Planning and designing figures—organizing, arranging, and visualizing data—is an important skill. 

Planning Figures

Figures are not merely collections of images that depict scientific data. They form narratives and tell stories just as much as the text does. In manuscripts today, each subheading-level finding or argument in the text roughly corresponds to one figure. That is, if the text opens with a paragraph showcasing data on how protein A modulates pathway A, the first figure should also cover that topic. 

Because figures form narratives just like text, they also should be planned like text—with particular emphasis on what information receives the most focus and how the information is ordered. Figures should include, at minimum, graphical data and/or schematics on the main arguments of the manuscript. It should also provide all of the necessary supporting evidence for those arguments, including key controls. Finally, it should clarify experimental strategies and methodologies when necessary to prevent the reader from having to hop back and forth between sections. 

Designing Figures

As figures have to be planned from an academic standpoint, they also have to be designed from a technical standpoint, particularly because figures in top journals can contain a dozen or more panels. The first question to consider is how many panels a figure should contain. Here, the answer is generally one panel for each argument made in the text. For example, in this article by Ruggero et al. published in Nature,¹ each piece of data corresponds to a figure panel:

On switching to a ketogenic diet, we observed an increase in AMPK phosphorylation and activity (Fig. 4a) in WT livers, as well as significantly increased P-eIF4E, while P-4EBP1 was downregulated (Fig. 4b and Extended Data Fig. 6a). Furthermore, pharmacologically inhibiting AMPK activity decreased P-MNK and P-eIF4E (Fig. 4c and Extended Data Fig. 6b). With a ketogenic diet, eIF4ES209A mice showed reduced blood BHB concentrations assessed after a short time (1 day) and showed a more than 50% reduction after long-term (2 months) feeding (Fig. 4d and Extended Data Fig. 6c–e).

The second question is how much information should go into each panel. The key here is to provide enough information to support the claim made in the text, but not go too far beyond that. Because each panel explores a single claim, one or two graphs/diagrams should be sufficient. To go back to Ruggero et al.’s publication, most of their panels contain graphs showcasing simple comparisons between two experimental groups. Extra supporting information that could provide context but is not essential to asserting the validity of the claim are placed in supplementary or extended data sets for the reader to explore at their own leisure.

Polishing Figures

Precision is critical when communicating scientific information. For figures, precision means not only technical accuracy, but also visual consistency. Attention to detail is vital to how effective a figure is at communicating information. For example, font sizes, faces, and styles should be deployed consistently across all figures. Just as in writing, how text is presented in figures conveys information to the audience, so scientists need to ensure that they are not saying something that they do not mean to say. Similarly, color usage should be consistent across all figures. If data from one experimental group is colored red in figure 1, it should not be blue in figure 2 and then green in figure 3. 

Beyond that, scientists need to take care to not draw attention away from their data, as the human eye has a keen sense for picking out discrepancies and is easily distracted. This means making sure graphs, images, labels, and panels are aligned correctly. It also means making sure that all figures have sufficient resolution for legibility and interpretation even when resized—something that is not always the case as researchers are not typically well-versed in graphic design concepts. Finally, it means arranging panels in an easy-to-follow manner, usually left to right, and using annotations to keep their eyes on the figure itself rather than darting to the text or the legend. These annotations can be as simple as including the P-value rather than using an asterisk, or using arrows or circles to highlight important areas of an image.

Acquiring an Eye for Design

Broadly speaking, scientific figures are not usually considered as something that falls within the graphic design umbrella. But applying design principles—as well as a skilled and trained designer—can help improve a figure’s aesthetic and clarity. As science and scientific data becomes more complex, scientists need to be aware of what they can do to make their figures more accessible to their audience.

Looking for some help putting together a manuscript, a figure, a poster, or anything else? Scientific Services may have the professional help that you’re looking for.