Improving Comprehension of Large Taxonomic Graphs

Taxonomic trees have been used for decades to visualize taxonomies; however, taxonomies involving many taxa may be difficult to interpret due to problems related to over- plotting. This problem can manifest itself in metagenomics studies, where the set of detected taxa can have a cardinality in the hundreds or thousands. We present a method by which a phylogenetic tree's complexity may be reduced by removing nodes with little support or with trivial out-degree (as defined by the user). A diffusive model is then used to color the nodes based on their taxonomic proximity; the color of ancestor nodes are a mixture of their descendents. This method results in compact taxonomic trees whose color gradually diffuses to white at the root. Our second example application of the technique is a more generalized structure of pedigrees. We show that related taxa can be easily located by reducing the complexity of the graph via pruning and coloring of related vertices.