ConstellationMap (v1.1)
Felix F. Wu, Yan Ni Tan · Zenodo (CERN European Organization for Nuclear Research) · 2015
ConstellationMap is a GenePattern module (http://www.genepattern.org/) that helps leverage the full power of a gene set enrichment analysis by identifying commonalities between high-scoring gene sets and mapping their relationships. ConstellationMap visualizes in an interactive web application the enrichment profile similarity and gene member overlap of gene sets, which are positively or negatively enriched in relation to a phenotype in a two-class or continuous-class comparison. It uses provided enrichment scores to estimate normalized mutual information (NMI) scores and project top scoring sets onto a circular plot with the following features: (1) Gene sets are represented as nodes with radial distance to the center reflecting positive or negative enrichment in the phenotype, whichever direction is specified. Higher associations for the correlation direction are plotted closer to the center. (2) Member gene overlaps between sets are represented as edges connecting nodes with thickness proportional to degree of overlap. (3) The angular distance between nodes is relatively proportional to the similarity of their enrichment profiles, i.e. more similar enrichment patterns have a smaller angular distance. Web interactive features allow export of selected overlapping gene symbols for annotation in DAVID, MSigDB, and GeneMania. Click on the output file Visualizer.html to open the interactive ConstellationMap plot from the Jobs Tab. Use mouse click on the web visualization to select a node or an edge. Use shift+click, or click-drag to select a group of features. Selection displays gene set names, number of genes in the set, number of sets used to compute overlap, and overlapping member gene symbols. Adjust number of nodes in highlighted area for which to display the maximum number of overlapping gene symbols. In addition to the web application, ConstellationMap outputs two static plots: (1) a positive red to negative blue heat map of per sample enrichment of the gene sets ranked by NMI scores as well as Area Under the Curve (AUC) and t-test metrics along with corresponding p-values, and (2) a constellation map marking the phenotype of interest in the center in red, concentric contour arcs marking association to the phenotype (as measured by NMI), nodes as numbered open circles, overlap as green lines, and a key of the numbered gene sets.