Sample Size for Microbiome Experiments

George Savva

Choosing the number of experimental units (i.e. the ‘sample size’) is an important part of any research study design. The larger the sample size, the more precise the estimates from your study will be, and the study will have more power to detect smaller effects.

Having too many samples has ethical and resource implications, so you may need to trade off this precision against ethics and cost.

However, under-powered studies are one of the main causes of irreproducibility in bioscience because they lead to high false positive rates  and a low chance of detecting true effects. Funding bodies are increasingly concerned with the detail of sample size calculations, particularly for animal and human studies, before they will fund research. Ethics committees also scrutinise sample size calculations, and reporting guidelines for both animal and human studies require that the rationale for determining study sample size is included in reports.

Having said this, sample size calculation is not an exact science, since it relies on many assumptions regarding the distribution of data and expected effects; and an assumed analysis plan that may not ultimately be the one that is followed.

In practice, a sample size calculation is required to (i) ensure that you have thought through the design and analysis of your study in relation to its aims and (ii) that your study is large enough to have a good chance of answering the question that it sets out to answer, to the level of certainty required, under a set of realistic assumptions about how the data might turn out.

The literature on power and sample size determination in microbiome studies is extremely limited (Appendix A). Although some microbiome-specific tools exist, general power and sample size tools can often be applied successfully, as well as ideas from studies based on gene expression datasets which often have a similar structure and aims.

This document sets out possible approaches to sample size calculation for microbiome studies and issues to be aware of. If you are not confident in calculating a sample size for your study, then always seek help from a statistician.

2 Sample Size for Microbiome Experiments v1.0 [pdf]

Creative Commons License
Quadram Institute Best Practice in Microbiome Research: Sample Size for a Microbiome Experiment v1.0 by George M Savva is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.