DBTree: Very large phylogenies in portable databases

Abstract

Growing numbers of large phylogenetic syntheses are being published. Sometimes as part of a hypothesis testing framework, sometimes to present novel methods of phylogenetic inference, and sometimes as a snapshot of the diversity within a database. Commonly used methods to reuse these trees in scripting environments have their limitations. I present a toolkit that transforms data presented in the most commonly used format for such trees into a database schema that facilitates quick topological queries. Specifically, the need for recursive traversal commonly presented by schemata based on adjacency lists is largely obviated. This is accomplished by computing pre- and post-order indexes and node heights on the topology as it is being ingested. The resulting toolkit provides several command line tools to do the transformation and to extract subtrees from the resulting database files. In addition, reusable library code with object–relational mappings for programmatic access is provided. To demonstrate the utility of the general approach I also provide database files for trees published by Open Tree of Life, Greengenes, D-PLACE, PhyloTree, the NCBI taxonomy and a recent estimate of plant phylogeny. The database files that the toolkit produces are highly portable (either as SQLite or tabular text) and can readily be queried, for example, in the R environment. Programming languages with mature frameworks for object-relational mapping and phylogenetic tree analysis, such as Python, can use these facilities to make much larger phylogenies conveniently accessible to researcher programmers.