Figure 7.
Sparsity and Polygenicity of Gene Expression in African Americans
We characterized the sparsity or polygenicity of gene expression traits by using a Bayesian sparse linear mixed model (BSLMM) analysis in the GTEx African American (AA) skeletal-muscle data. We estimated the proportion of variance in gene expression that can be explained by sparse effects () and the proportion of variance in gene expression that can be explained by sparse effects and random effects together (), the latter of which is most equivalent to our LMM-based . In the genes analyzed, estimated values that were significant at p value < 0.05 were defined as nominally significant estimates.
A. The comparison of and the estimate from the BSLMM. Genes with a large and a large estimate are likely to have highly sparse local genetic architecture.
B. The comparison of from GCTA and the estimate from the BSLMM showing the independence of the two components.