Zoomable Heatmaps: Improving Manhattan Plots to Compare Multiple Genome-Wide Studies

Making comparisons between genome-wide association studies is a crucial practice in genetics. Despite this, conventional visualizations used in these studies tend to suffer from readability issues. These issues include, for instance, overcrowded data points and difficulties in supporting comparisons between different distributions. One of the most commonly used visualizations for this purpose, the traditional Manhattan plot, is visually optimized for comparing peaks only within one distribution of data points. In genetics research, however, there is often a need to compare multiple distributions. In this paper, we present a novel interaction model that relies on the use of zoomable heatmaps. This interaction model aims to support users in preliminary and straightforward cross-distribution comparisons through semantic zooming, heatmaps, and the overview+detail approach. All these interaction methods have their origin in map-based user interfaces and visualizations.