A comprehensive reference dataset will also help researchers make rational decisions about sample size by enabling power calculations, which can greatly impact the utility of a study [34]. Such a dataset is also crucially necessary to support characterization of the effect sizes of variables (e.g., antibiotic use). Within the microbiome field, effect size for many variables of interest is not yet well understood, and many that are important in diseases with complex etiologies such as autism [10] are likely to be small. Well-characterized references offer the possibility for a researcher to expand their dataset by pulling reference samples to augment their own [29], particularly when meta-analysis (i.e., combination of summarized data from multiple studies) is taken into consideration during the design phase for a study.
