Genetic QTL Mapping of Gut Microbiome and Lipidome

Genetic QTL mapping was performed using the R/qtl2 (v0.24) package^{78 (link)} which fit a linear mixed effect model that included accounting for overall genetic relationship with a random effect, that is, kinship effect. The leave one chromosome out (LOCO) method was used, which accounts for population structure without reducing QTL mapping power. For each gut microbiome trait and caecal lipidome traits, sex, days on diet and mouse cohort (wave) were used as additive covariates as described previously^{13 (link)}. For gut microbiome traits and caecal lipidome traits, normalized abundance/coverage was transformed to normal quantiles. The mapping statistic reported was the log₁₀ likelihood ratio (LOD score). The QTL support interval was defined using the 95% Bayesian confidence interval^{78 (link)}. Significance thresholds for QTL were determined by permutation analysis (n = 1,000). We included 2,803 gut microbiome function traits, 197 gut microbiome taxon traits and 3,384 caecal lipid feature traits for our QTL mapping. The reported genome-wide P values were not adjusted for the multiple phenotypes to avoid overly declaring QTL in the initial analysis. We used genome-wide P < 0.05 for significant QTL and used genome-wide P < 0.2 to find concordant QTL mapping and hotspots.