Scientific writing is about communication
You’ve spent a long time thinking about a particular problem, you’ve done your own investigation and experimentation, and you’ve found something worth sharing with others. To share that discovery, you need to communicate it effectively.
There are many ways to communicate with others
Conferences, seminars, and blog posts are all methods for sharing your work. Many of these methods make use of visuals to help the audience understand your findings. Since ideas are complicated, it’s easy for visuals to be complicated. If your audience has a difficult time interpreting the visuals that are meant to help them understand, however, you’re undercutting your work and its message.
Complicated visuals are a barrier to good communication
Making visuals for your work becomes a delicate balancing act. You want to accurately represent your findings with visuals that inform your audience about the key results, discoveries, and insights without misrepresenting or overstating your findings. You should make then as simple as possible without misrepresenting them, but no simpler. This optimization problem requires as much creativity as you can muster and it helps to take advantage of as many tools as you can.
Colour is an effective visual communication tool
Using an eye-grabbing colour can direct the audience’s attention, but too much of it may make it hard for them to know where to look. Using the same colour repeatedly can help reinforce a theme you’re trying to emphasize. Using a distinct colour makes the same figure equally understandable on white pages and in dimly-lit lecture halls.
Using colour is easy to mess up
Artists, designers, and software developers know this all too well. Scientists can’t escape from these difficulties, either. Consider this article discussing the many variants on a simple palette that are required to give a program a smooth and realistic feeling while still communicating the many things the program can do to the user without overwhelming them. There are many decisions made at each step of this design process, each with their own tradeoffs.
Picking colours isn’t a science
There is a lot of science about colour, though. There are also a lot of tools to help you do it without needing to go to art school. Here is a recent perspective about how colour is misused in scientific communication. Importantly, this article offers a guide on how to pick colours effectively. Fittingly, it can be summarized with this figure:
So simple, right?
The previous article focuses on colour as a way to represent data. There is one key factor that, in my opinion, this article is missing.
Colour can, and should, be used to represent experimental design
Any paper will likely make comparisons between groups of observations, like a control group and a treatment group. Experiments can be more complicated than this, but using colour in this manner helps the reader keep track of what is actually being shown, not just the values that are being shown.
From this panel, you can immediately recognize the factors in their experiments. They are interested in active promoters, active enhancers, and primed enhancers in 4 tissue types across 10 animals. These 3 distinct colours are repeatedly used throughout the paper.
Here is another panel from a different figure:
You don’t need to think that hard about what is being compared because you can clearly see it. This paper effectively uses colour and icons to represent what is being compared, not just the values from their analyses.
Look for tools that others have built to make colour easy to work with
Here are a few resources that I repeatedly use for figuring out what colours to use for scientific communication.
- pastel: a command line tool for manipulating colours
- Affinity Designer: a professional vector graphics design tool. Similar to Adobe Illustrator.
- R color palettes: an R package for making colour maps2
- colorblindr: an R package to test how your ggplot colours will look to people with colour deficiencies3
- I Want Hue: using data science methods to make colour paletts for data science visuals
- Color Brewer: colour suggestions for cartography
Look for examples that inspire you when designing figures and documents
- Nature Reviews Genetics is a journal specializing in review papers for fields related to genetics. Their graphic designers use consistent schematics and colour schemes for communicating ideas across multiple articles. This consistency lets readers pick up articles from years ago and still find the same visual elements as today. This makes connecting ideas and following trends over time much easier.
- Quanta Magazine is an excellent science journalism magazine. It primarily focuses on math and physics, but their graphics designers produce incredible visuals for their articles.
- FiveThirtyEight is a news organization focusing on politics and sports. They repeatedly use orange, green, and purple in their plots and offer a unique style to communicate with a broad audience.
- The Pudding is a digital publication that explains ideas debated in culture with visual essays. They have incredible articles that show how to use the internet as an interactive medium for telling stories.
- The Markup is another news organization, similar to The Pudding, that focuses on technology.
The science in this paper was okay, as far as I could tell. But the paper was very confusing in a few parts (Figure 2C), and didn’t spend much time on the thing in the title of the paper. It’s a bit misleading, as far as titles go. I wonder if that was done because it’s catchy. (Update 2021-02-18: this comment was in regards to the preprint. The article has now been published in Genome Biology, and the link above is updated). ↩
I mostly just look at the README, though. ↩