Bayesian Modeling of Rumen Microbiota Composition Using Host Genomic Markers and Microbial Abundances

In dairy cattle, the rumen microbiota plays a crucial role in the digestion of fiber and converting it to high-quality protein. Several studies have shown that host genetics can influence the composition of the rumen microbiome in dairy cattle. However, a deep understanding of microbial functions (MFs) associations and the direct effects of single-nucleotide polymorphisms (SNPs) on MF abundances remains limited. In this study, a Bayesian sparse graphical model was used to jointly estimate SNP-MF and MF-MF associations using the 16s predicted metabolic pathways (16s-PMP) and genotype data from 338 Holstein dairy cattle. Our analysis identified two significant SNPs associated with bacterial MF abundances involved in RIG-I-like receptor signaling and Renin angiotensin system. Genome-wide association studies (GWAS) supported these findings by identifying SNPs related to the same KEGG pathway identified through the Bayesian model. In terms of MF-MF interactions, we identified several hub nodes such as glycolysis/gluconeogenesis and Lipid metabolism pathways, which play a role in methane emission. Overall, our results suggest that the effects of host genetic on microbiota composition are modest. This may partly reflect the limited statistical power resulting from the sample size and model complexity. However, our graphical model can provide a better understanding of direct SNP-MF and MF-MF associations within the microbiome. This may help to identify specific microbial targets with a potential role in methane emission.