The linear mixed model underlying variancePartition allows the effect of one variable to depend on the value of another variable. Statistically, this is called a varying coefficient model [31]. This analysis examines the expression variation across individuals within multiple cell types, or another subset of the data. A given sample is only from one cell type, so this analysis asks a question about a subset of the data. The data is implicitly divided into subsets based on cell type and variation explained by individual is evaluated within each subset. This subsetting means that the variance fractions no longer sum to 1, but the model still allows ranking of dimensions of variation based on genome-wide contribution to variance and enables analysis of gene-level results. See the Additional file 1 for more details.
