Fig. 4. Mediation effects of ben-1 expression on C. elegans resistance to albendazole.

a Mediation estimates (y-axis) calculated as the indirect effect that differences in expression of each gene play in the overall phenotype are plotted against the genomic position of the eQTL (x-axis) on chromosome II. The horizontal gray line represents the 99th percentile of the distribution of mediation estimates. Significant mediators are colored other than gray by their genes as shown in the legend. b The correlation of animal length (y-axis) to raw ben-1 expression and to ben-1 expression regressed by variation at ben-1 on y-axis. The correlation coefficient r and the p values using the two-sided Pearson’s correlation tests were indicated at the bottom. Strains are colored by the type of their genetic variants in ben-1. Strains without identified variants are colored gray. c Significance at the pseudo variant marker of 25,837 GWA mappings. Each point represents a GWA mapping that is plotted with its −log₁₀(p) value (y-axis) at the pseudo variant marker (III: 3,539,640) against the genomic locations (x-axis) of the transcript of which the expression was used in regression for animal length. Points for traits regressed by expression of transcripts identified as significant mediators are colored as in a. The orange horizontal line represents the significance of the pseudo variant marker using the raw animal length of 167 strains (Supplementary Fig. 6). GWA mapping results of 12 traits regressed by expression of mitochondrial genes were excluded but all with significance close to the horizontal line. We used the GWAS() function in the R package rrBLUP⁸⁷ to perform the genome-wide mapping with the EMMA algorithm³² (see “Methods”).