This text was initially written for the GSA Early Career Leadership Newsletter and was since then modified to fit my blog format
In the last decade, most journals have moved to online publication only, which means charging an extra fee for color figures is no longer standard. A good set of colors can make any dull and aggressively ugly bar chart into something intriguing and can highlight your key message. Here are a few pieces of advice I learned in my data visualization journey:
Never go with the default. All software that helps you plot your data come with overused and abused preset colors. Everyone knows them and secretly dislikes them. Change those, to show an inch of involvement in making the figure. Additionally, think about color blind people, do you always need to go with the old green vs red contrast?
What is your key message? For each figure in your article, you need to identify the core message you want to convey and select which element will be the most important to highlight.
Choose a color palette. Once you know what to focus on, think about a color palette. Ideally, 4 to 5 colors are enough to make clear distinctions. If you need more, try and find ways to cluster your variables. Can the less important genes from the same chromosome be colored in different shades of orange?
Some colors work together and others don’t. Plenty of websites can help you with this, check Adobe Color Wheel, Coolors and I want hue. For your figure: Choose a set of bright, saturated, but not extreme colors for your significant parameters and contrast them against soft, unsaturated, ones.
Be consistent. Once you have a color palette and design style, stick to it. Be consistent across all the figures in your publication. The color of parameter A in figure 1 should stay the same in the subsequent ones. Choosing color is a science, and some color works better with others, many websites can help you find colors that fit together
Do what you like. This is the final advice, it is stressful enough in academia without having to worry about something else. Nothing is more easily forgotten than panel 2b from a paper or slide 32 from your presentation. Your science is what will drive you forward, bad data even disguised with pretty makeup will not give you high-impact publication. So relax, use colors you like, and try to remember those few tips when you are close to publishing.