High Throughput Reproducible Literate Phylogenetic Analysis

Abstract

We present a holistic approach from a literate programming perspective to frame and solve systems biology problems. In particular, given the large data-sets required for answering questions relating to evolutionary histories we focus on the generalization and workflow required on a typical SLURM or PBS TORQUE queue driven high performance computing cluster. We demonstrate how to leverage multiple CLI tools compiled for efficient use in a portable manner on heterogeneous computational resources and further demonstrating the use of R to generate literate data-driven plots and analysis. High Performance Computing cluster (HPC) bottlenecks and installation barriers are also discussed and mitigation strategies are developed. As a concrete example we demonstrate the estimation of a phylogenetic tree, used to pose and answer questions on evolutionary lineages. In this manner, a generalized approach which can be used for systems biology is elucidated for manipulating phylogenetic data, including its validation, multiple sequence alignment, tree estimation through different models and reproduction.