Fig. 4From: A sparse Bayesian factor model for the construction of gene co-expression networks from single-cell RNA sequencing count dataProperties of PPD estimates from a sample of nine genes in the MMC dataset. Genes were selected based on percentiles (10th through 90th) of average gene expression. a Violin plots of estimated log(variance/mean) for each gene across all PPDs. b Violin plots of estimated gene-specific proportion of zeros across all PPDs. The blue stars represent the true values from the MMC datasetBack to article page