Biological and technical variability in mouse microbiome analysis and implications for sample size determination

Abstract

BackgroundThe gut microbiome (GM) affects host growth and development, behavior, and disease susceptibility. Biomedical research investigating the mechanisms by which the GM influences host phenotypes often involves collecting single fecal samples from laboratory mice. Many environmental factors can affect the composition of the GM in mice and while efforts are made to minimize these sources of variation, biological variation at the cage or individual mouse level and technical variation from 16S rRNA library preparation exist and may influence microbiome outcomes. Here we employed a hierarchical fecal sampling strategy to 1) quantify the effect size of biological and technical variation and 2) provide practical guidance for the development of microbiome studies involving laboratory mice. ResultsWe found that while biological and technical sources of variation contribute significant variability to microbiome alpha and beta diversity outcomes but their effect size is 3- to 30-times lower than that of the experimental variable in the context of an experimental group with high intergroup variability. After quantifying variability of alpha diversity metrics at the technical and biological levels, we then simulated whether sequencing multiple fecal samples from individual mice could improve effect size in a two-group experimental design. Collecting five fecal samples per mouse increased effect size achieving the maximum 5% reduction in the required number of animals per group. While reducing the number of animals required, sequencing costs were dramatically increased. ConclusionsOur data suggest that the effect size of biological and technical factors may contribute appreciable variability to an experimental paradigm with relatively low mean differences. Additionally, repeated sampling improves statistical power however, its application is likely impractical given the increased sequencing costs.