IDENTIFICATION OF DYNAMIC MICROBIAL SIGNATURES IN LONGITUDINAL STUDIES

The study of microbiome dynamics is key for unveiling the role of the microbiome in human health. Addressing the compositional structure of microbiome data is particularly critical in longitudinal studies where compositions measured at different times can yield to different subcompositions. We propose a new compositional data analysis (CoDA) algorithm for inferring dynamic microbial signatures. The algorithm performs penalized regression over the summary of the log-ratio trajectories (the area under these trajectories) and the inferred microbial signature is expressed as a log-contrast model. Graphical representations of the results are provided to facilitate the interpretation of the analysis: plot of the log-ratio trajectories, plot of the signature and plot of the prediction accuracy of the model. The new proposal is illustrated with data on the developing microbiome of infants. The algorithm is implemented in the R package "code4microbiome" (https://cran.r-project.org/web/packages/coda4microbiome/) that is accompanied with a vignette with a detailed description of the functions. The website of the project contains several tutorials: https://malucalle.github.io/coda4microbiome/