Effect of Internal Standard Normalization of Microbiome Data on Outcomes of a Controlled Feeding Study and a Longitudinal Study in a Multiethnic Cohort

Standardization would benefit the interpretability of human microbiome data because unintended variability can be introduced at each level of data production and processing. One way to bring standardization to microbiome studies is with internal standards (IS). In three microbiome sequencing methods—16S rRNA gene sequencing, and metagenomic and metatranscriptomic sequencing—this standardization can involve the addition of nucleic acid sequences into a sample. We assessed how internal standards would change the interpretation of the data in a controlled feeding study and longitudinal study in a multiethnic cohort. We compared both coefficients of variation (CV) and intraclass correlation coefficients (ICC) for both IS-normalized data and non-normalized data for both studies. The effect of IS-normalization on the ICCs and CVs was inconsistent in our results and did not improve data interpretation.