Exploring the Functional Landscapes of Gene Sets with Interactive Multidimensional Scaling

We present a multidimensional-scaling algorithm and novel visualization tool for interactive exploration of the functional roles of genes in a gene set. We reference a popular database of gene attributes to build a two-dimensional projection of a user’s set of query genes that reveals the functional structure of a set of genes as well the relationship of the query genes to the whole gene landscape, including approximate ‘enrichment’ analysis. We show that the projection effectively visualizes broad biologically meaningful structures in the gene landscape. We also show that the visualization and interaction methods provide users with better context and enable more fluid and rapid preliminary exploration than previously existing tools, demonstrated through two use cases. We discuss possibilities for improving the performance of the algorithm to capture more fine-grained relationships between genes.