Fig. 4: Interpretation of loadings.
From: Transcripts with high distal heritability mediate genetic effects on complex metabolic traits
A. Loadings across traits. Body weight and insulin resistance contributed the most to the composite trait. B. Phenotype scores across individuals. Individuals with large positive phenotype scores had higher body weight and insulin resistance than average. Individuals with large negative phenotype scores had lower body weight and insulin resistance than average. C. Distribution of transcript loadings in adipose tissue (purple). For transcripts with large positive loadings, higher expression was associated with higher phenotype scores. For transcripts with large negative loadings, higher expression was associated with lower phenotype scores. D. Distributions of loadings across tissues compared to null distributions. Shaded areas represent loadings that were more extreme than the null distribution. Numbers indicate how many transcripts had loadings above and below the extremes of the null. Transcripts in adipose tissue (purple) had the most extreme loadings indicating that transcripts in adipose tissue were the best mediators of the genetic effects on body weight and insulin resistance. E. Scatter plots showing correlations between composite vectors for the genome (GC), the transcriptome (TC), and the phenome (PC). The GC - TC association was significant (Linear regression R2  = 0.18; beta coefficient = 7.8 ± 0.86 standard error; t = 9.03; p < 2. 2−16). The TC - PC association was significant (Linear regression R2 = 0.62; beta coefficient = 3.1 ± 0.13 standard error; t = 24.4; p < 2. 2−16). There is no association between GC and PC (Linear regression R2 = 7.1 × 104; beta coefficient = 2.0 ± 3.8 standard error; t = 0.51; p = 0.61). This correlation structure is consistent with perfect mediation. Blue lines show lines of best fit. Source data are provided as a Source Data file.
