All hail reproducibility in microbiome research

At its core, microbiome research relies on complex and sophisticated statistical analyses of large datasets and their associated metadata (e.g., experimental parameters, sample characteristics). Heavy reliance on big data has presented new challenges in communicating the details of complex analyses in a manner sufficient for others to replicate analytical workflows. Reproducibility is a pillar of sound research, and scientific journals need to embrace transparency and make every effort to enable reproducibility through comprehensive and clear reporting of analytical approaches. In this issue of Microbiome, a report by Meadow et al. [1] on the microbial communities of classroom surfaces sets a new bar for thoroughness in the availability of data, metadata, and analytical resources (code and scripts). It is our hope that this paper will serve as a template for the clever use of publicly available resources and code repositories to enable fully reproducible microbiome research. “Scientific publications have at least two goals: (i) to announce a result and (ii) to convince readers that the result is correct... papers in experimental science should describe the results and provide a clear enough protocol to allow successful repetition and extension” [2]. Reproducibility and extension are only possible if: data is easily and freely accessible and delivered in format that adheres to international standards; and analysis workflows and scripts are embedded in the publication. Microbiome research is, by its nature, a multi-disciplinary endeavor where experimentalists often work with biostatisticians, mathematicians, computer scientists, or epidemiologists. At times, this multi-disciplinary character can result in a clash of scientific cultures with different approaches to openness, transparency and data release. For example, large sequence datasets and most importantly associated metadata have resulted from our work with epidemiologists [3,4]. However, the notion of releasing data and analysis scripts along with a publication has often been met with great surprise by our epidemiology colleagues. Now that