Structuring data analysis projects in the Open Science era with Kerblam!

Abstract

Background: Structuring data analysis projects, that is, defining the layout of files and folders needed to analyze data using existing tools and novel code, largely follows personal preferences. Open Science calls for more accessible, transparent and understandable research. We believe that Open Science principles can be applied to the way data analysis projects are structured.

Methods: We examine the structure of several data analysis project templates by analyzing project template repositories present in GitHub. Through visualization of the resulting consensus structure, we draw observations regarding how the ecosystem of project structures is shaped, and what salient characteristics it has.

Results: Project templates show little overlap, but many distinct practices can be highlighted. We take them into account with the wider Open Science philosophy to draw a few fundamental Design Principles to guide researchers when designing a project space. We present Kerblam!, a project management tool that can work with such a project structure to expedite data handling, execute workflow managers, and share the resulting workflow and analysis outputs with others.

Conclusions: We hope that, by following these principles and using Kerblam!, the landscape of data analysis projects can become more transparent, understandable, and ultimately useful to the wider community.