Archive for March, 2012
Making Beautiful Plots: Data Display Basics
You’ve labored at the bench and generated data that you’re about to meticulously analyze before preparing the results of your hypothesis-testing for presentation. In this post, we’ll discuss elements that factor into making beautiful (and consistent) displays of data. View our recent post on Analysis Consistency in Flow Cytometry for a discussion of broader themes relating to analysis consistency.
To summarize what will follow in short: make sure all of your data are on scale, accurately compensated, and make sure all your plots are well-labeled.
Choosing plot types, appropriate statistics, and telling the full story
There are a number of plot types that can help you tell your story in different, visually pleasing ways when used appropriately. Among the flashier ways to display data are heatmaps, histograms, and histogram overlays. These one-dimensional representations owe their appeal largely to their ability to convey an easy-to-understand message: “This population changed in X amount in Y condition.” Where this gets tricky is if you’re trying to describe a heterogeneous population. When deciding on a plot type to use to convey your story, you’ll want to make sure you’re telling the whole story, and not omitting important information about the behavior of subsets in the course of eliminating a dimension of data display. In Cytobank, you can mouse over a heatmap square to display the underlying dot plot, which will reveal another dimension of information of your data.

Figure 1. Example of a well-labeled figure using one- and two-dimensional representations.
Excerpted from Irish JM et al (2010) PNAS, 107(29):12747-54, Figure 1B.
(Click on the image for higher resolution)
(more…)

