Subtree extraction benchmarks

This data set demonstrates the difference in processing time between a DBTree-indexed version of a tree and a Newick version. For both versions, a varying number of randomly drawn tips, ranging from 10 to over 40k, is extracted as a subtree from the topology of the Open Tree of Life.
On small subsets, the difference is especially dramatic: where subtrees for dozens of taxa are extracted in less than a second from the DBTree version, the Newick tree takes over ten minutes to be processed. Much of the processing time for the latter is taken up simply by the Newick string parsing operation.
For large subsets, the difference is less dramatic. For example, for a subtree of over 40k taxa, the DBTree implementation is about twice as fast as the Newick/Dendropy implementation. However, the difference in processing time is now no longer due primarily to the Newick string parsing: the initial data reading step takes about 10 minutes, but the total operation takes over two hours.
For more information, see: https://github.com/rvosa/bio-phylo-forest-dbtree