Thumbnails:
List:
Year:
Category:
Session:
Poster:
Getting poster data...
Ryo Sakai, Raf Winand, Toni Verbeiren, Andrew Vande Moere, Jan Aerts (Department of Electrical Engineering (ESAT) STADIUS Center for Dynamical Systems, Signal Processing and Data Analytics, KU Leuven, 3001, Belgium. iMinds Medical IT, KU Leuven, 3001, Belgium)
A cluster heat map is a popular cluster analysis method with a wide range of applications, including visualization of adjacency matrices and gene expression profiles from high-throughput experiments. This method utilises agglomerative hierarchical clustering to reorder rows and columns of the input matrix based on the leaf orders from resulting dendrogram structures. Then, the reordered matrix is visualized as a heat map with dendrogram for rows and columns showing the cluster-subcluster relationships and the clustering process. Although this method is widely used and accepted, it has some shortcomings. First, the leaf order of a dendrogram does not reflect either the monotonic order in which clusters are merged or the nested cluster relationships. Second, the distance between nodes or subclusters is encoded only in the dendrogram and it is not conveyed in the heat map visualization. To address these shortcomings, we developed new leaf ordering methods and a novel visual encoding for heatmap, implemented as R packages (“dendsort” and “gapmap” respectively).