Fig. 1
From: Pitfalls and opportunities for applying latent variables in single-cell eQTL analyses
Correlation among inferred PEER factors and global intra-individual mean–variance dependence. A Pair-wise correlation plot among the first 10 PEER factors generated from single-cell expression in CD4NC (Naive CD4) cells without any quality control (option #1). The upper triangle panel shows the pair-wise estimates of Pearson’s correlation, and the bottom triangle panel shows the pair-wise scatter plot between the PEER factors. The diagonal panel shows the distribution of each PEER factor. The significance of the correlation test is annotated by * p-value ≤ 0.05, ** ≤ 0.01, *** ≤ 0.001. B Diagnostic plot of the factor weights without further quality control on the pseudo-bulk matrix (option #1, upper panel) and option #11 QC (lower panel). C Relationship between intra-individual pseudo-bulk mean and Fano factor per gene. Both axes are Log10 transformed. The color of the dots indicates the proportion of zero expression across individuals (\({\pi }_{0}\)) for each gene. D Scatter plot of first 10 PEER factors generated from all genes against those from top 2000 highly variable genes (option #11 vs option #12)