Fig. 2

Varying degrees of parental bias among imprinted genes detected in LCLs, WB, and GTEx. Each point represents the PatRatio (the mean fraction of reads transcribed from the paternal allele) in each informative individual per gene, with the point size indicating total read depth over all heterozygous transcribed SNVs in that sample. Genes are ordered left to right by increasing mean PatRatio. The upper panel shows stranded data from LCLs, while the lower panel shows unstranded data from WB samples. Note that due to the very low read depth in some genes/individuals, several genes showed highly variable PatRatios within the population. A small x- and y-axis jitter was added to reduce overplotting effects. Genes shown in black were significant (FDR < 0.1), while those in red did not pass this statistical threshold for significance. The figure is divided into three panels: left, middle, and right panel. Genes in the middle panel were found significant in LCL and/or WB and reported as putatively imprinted in GTEx [11]; genes shown in the left panel were found significant in LCL and/or WB but not reported in GTEx; and genes shown in the right panel represent those reported as putatively imprinted in GTEx but were not identified as showing significant evidence of imprinting in either LCL or WB. Some genes in the right panel such as DLK1, MEG9, THEGL, DIRAS3, PWRN1, and NDN show evidence of parental expression bias, but the limited number of informative samples meant we did not consider these in our formal analysis. For genes with multiple UGFs (Additional file 3), we plot paternal ratios for the UGF with the most significant p value