Fig. 5: Estimates of cell type fixed effect and cell type-specific variance for specific genes in REML with CTP data. | Nature Communications

Fig. 5: Estimates of cell type fixed effect and cell type-specific variance for specific genes in REML with CTP data.

From: A robust model for cell type-specific interindividual variation in single-cell RNA sequencing data

Fig. 5: Estimates of cell type fixed effect and cell type-specific variance for specific genes in REML with CTP data.

Four genes were chosen as examples. POU5F1 exhibited the strongest signal of variance differentiation among all genes; NDUFB4 exhibited the strongest signal of variance differentiation among genes without signals of mean differentiation (\(p > 0.01\)); MIXL1 exhibited the strongest signal of mean differentiation among all genes; and EOMES exhibited the strongest signal of mean differentiation among genes without signals of variance differentiation (\(p > 0.01\)). The violin plot represents the distribution of cell type-specific pseudobulk after standardizing overall pseudobulk to mean 0 and variance 1; circles indicate estimated cell type fixed effects; the length of the dashed line indicates twice the sum of homogeneous variance shared across cell types and cell type-specific variance. \(P\) values for testing mean differentiation and variance differentiation were calculated using the jackknife-based Wald test with CTP data from 94 individuals.

Back to article page