Towards Modular Components for Phylogenetic Estimation

Workflows have been a staple of the biological sciences for a while now, but in the context of generating more physically relevant insights it is necessary to have reusable modular components. Taking inspiration from various sources like the Electronic Structure Library initiative for ab-initio calculations or the Atomic Simulation Environment, we describe a class hierarchy for efficiently working with phylogenetic analyses and queries. In particular, we shall discuss the reference implementations of a python library which provides object oriented structures for evolutionary studies while also being able to interact with workflow engines which can then scale on high performance computing systems. Software design for these components is non-trivial due to the fact that many of the high compute requirements have separate structures, e.g. for exploring probability modes in Bayesian phylogenetic tree estimation. By demonstrating a common language and framework for expressing these constraints we will also tie into the key component of biological relevance, visualization, and the library oriented design scales to cross-language workflows linking C++ / Fortran / Python and R components and MPI variants of the same.