A critical consideration in any clinical study is power analysis, yet it has been difficult to perform such analyses for microbiome studies because the effect sizes of different disorders are unknown. Fortunately, several larger cohort studies, including but not limited to the Human Microbiome Project, now allow us to identify effect sizes for differences among ages and populations, and differences associated with obesity, IBD, and other disorders.
The original Evident demo was a web-based software tool that can be found here.
Evident is comprised of the following steps:
- clean your metadata file to only contain categorical metadata columns that are not derived from other columns, for example: barcode with sample name or any numeric column should be binned
- find your alpha and beta diversity calculation files, in classic qiime 1 format, based on which one you want to estimate the effect size
- run evident effect-size using your clean metadata file and your alpha and beta files; note that you need to define which none values the effect size should ignore (via --na_values) e.g. 'NA', ' ', 'None', 'Not Applicable'
- run the effect size summaries evident summarize