Abstract
Making scientific analyses reproducible, well documented, and easily shareable is crucial to maximizing their impact and ensuring that others can build on them. However, accomplishing these goals is not easy, requiring careful attention to organization, workflow, and familiarity with tools that are not a regular part of every scientist's toolbox. We have developed an R package, workflowr, to help all scientists, regardless of background, overcome these challenges. Workflowr aims to instill a particular "workflow" - a sequence of steps to be repeated and integrated into research practice - that helps make projects more reproducible and accessible.This workflow integrates four key elements: (1) version control (via Git); (2) literate programming (via R Markdown); (3) automatic checks and safeguards that improve code reproducibility; and (4) sharing code and results via a browsable website. These features exploit powerful existing tools, whose mastery would take considerable study. However, the workflowr interface is simple enough that novice users can quickly enjoy its many benefits. By simply following the workflowr"workflow", R users can create projects whose results, figures, and development history are easily accessible on a static website - thereby conveniently shareable with collaborators by sending them a URL - and accompanied by source code and reproducibility safeguards. The workflowr R package is open source and available on CRAN, with full documentation and source code available at https://github.com/jdblischak/workflowr.
Highlights
A central tenet of the scientific method is that results should be independently verifiable — and, ideally, extendable — by other researchers
In the remainder of this article, we describe the workflowr interface, explain its design, and give examples illustrating how workflowr is used in practice
Wflow_build() summarizes the results of these reproducibility safeguards in a report at the top of the webpage, along with additional “reproducibility checks”, which alert the user to potential reproducibility issues, such as changes that were not committed to the project development history, and the use of absolute file paths (Figure 2)
Summary
Hickey , Walter and Eliza Hall Institute of Medical Research, Parkville, Australia. Any reports and responses or comments on the article can be found at the end of the article. Open science, workflow, R, interactive programming, literate programming, version control. This article is included in the RPackage gateway
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
