ReproPhylo: An environment for reproducible Phylogenomics

Amir Szitenberg*, Max John, Mark L. Blaxter, David H. Lunt

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

Abstract

The reproducibility of experiments is key to the scientific process, and particularly necessary for accurate reporting of analyses in data-rich fields such as phylogenomics. We present ReproPhylo, a phylogenomic analysis environment developed to ensure experimental reproducibility, to facilitate the handling of large-scale data, and to assist methodological experimentation. Reproducibility, and instantaneous repeatability, is built in to the ReproPhylo system and does not require user intervention or configuration because it stores the experimental workflow as a single, serialized Python object containing explicit provenance and environment information. This 'single file' approach ensures the persistence of provenance across iterations of the analysis, with changes automatically managed by the version control program Git. This file, along with a Git repository, are the primary reproducibility outputs of the program. In addition, ReproPhylo produces an extensive human-readable report and generates a comprehensive experimental archive file, both of which are suitable for submission with publications. The system facilitates thorough experimental exploration of both parameters and data. ReproPhylo is a platform independent CC0 Python module and is easily installed as a Docker image or a WinPython self-sufficient package, with a Jupyter Notebook GUI, or as a slimmer version in a Galaxy distribution.

Original languageEnglish
Article number1004447
Number of pages13
JournalPLoS Computational Biology
Volume11
Issue number9
DOIs
Publication statusPublished - 3 Sep 2015

Keywords

  • protein-sequence alignments
  • phylogenetic analysis
  • high-throughput
  • evolution
  • system
  • galaxy
  • information
  • generation
  • standard
  • workflow

Cite this