An integrated strain-level analytic pipeline utilizing longitudinal metagenomic data

The development of sequencing technology and analytic tools have advanced our insights into the complexity of microbiome. Since different strains within species may display great phenotypic variability, studying within-species variations enhances the understanding of microbial biological processes. However, most existing methods for strain-level analysis do not allow for the simultaneous interrogation of strain proportions and genome-wide variants in longitudinal metagenomic samples. In this study, we introduce LongStrain, an integrated pipeline for the analysis of metagenomic data from individuals with longitudinal or repeated samples. Our algorithm improves the efficiency and accuracy of strain identification by jointly modeling the strain proportion and genomic variants in combined multiple samples within individuals. With simulation studies of a microbial community and single species, we show that LongStrain is superior to three extensively used methods in variant calling and proportion estimation. Furthermore, we illustrate the potential applications of LongStrain in the real data analysis of The Environmental Determinants of Diabetes in the Young (TEDDY) study and a gastric intestinal metaplasia microbiome study. We investigate the association between the dynamic change of strain proportions and early life events, such as birth delivery mode, antibiotic treatment, and weaning. By joint analysis of phylogeny and strain transition, we also identify a subspecies clade of Bifidobacterium longum which is significantly correlated with breastfeeding.