parafac4microbiome: Exploratory analysis of longitudinal microbiome data using Parallel Factor Analysis

Studies investigating microbial temporal dynamics are increasingly common, leveraging longitudinal designs that collect microbial abundance data across multiple time points from the same subjects. Traditional exploratory approaches like Principal Component Analysis (PCA) fail to fully utilize this structure. By organizing data as a three-way array - subjects as rows, microbial abundances as columns, and time points as the third dimension - multi-way methods such as Parallel Factor Analysis (PARAFAC) can better capture temporal and structural patterns. This study demonstrates Parallel Factor Analysis (PARAFAC) as a method to explore longitudinal microbiome data using three exemplary studies. In the first example, a long time series of in vitro microbiomes, PARAFAC identifies primary time-resolved variations. The second example, a longitudinal infant gut microbiome study, shows that PARAFAC can distinguish subject groups and enhance comparative analysis, even with moderate missing data. In the third example, a gingivitis intervention study of the oral microbiome, PARAFAC enables the identification of microbial subcommunities of interest through post-hoc clustering. These examples highlight PARAFAC's broad applicability for analyzing longitudinal microbiome data across diverse environments. The approach is implemented in the R package parafac4microbiome, available on CRAN, providing researchers with accessible tools for similar analyses.