Fig. 3: Mechanisms of RNA abundance regulation via 3′ UTR.
a Relative activity scores of functional variants that disrupt miRNA target sites. HEK293: N = 6189 unique variant-target pairs. HeLa: N = 7569 unique variant-target pairs. P-values are obtained by a one-tailed Wilcoxon rank sum test. b Functional variants in PLIN4 (HEK293) and LDHD (HeLa) disrupt the target sites for miR-34b-3p (top) and miR-3180-5p (bottom), respectively. MapUTR activity scores are shown in the middle. N = 3 from biologically independent replicates. Created with BioRender.com. c Functional MapUTR variants significantly alter RBP binding (See also Supplementary Fig. 8, 9). X-axis shows the binding score difference between the reference and alternative alleles predicted by DeepRiPe. P-values were calculated using a two-tailed Kolmogorov-Smirnov test. The p-values indicated in the plot are the maximum p-values obtained from the two comparisons (functional versus nonfunctional variants [HEK293: p < 2.2 × 10−16, HeLa: p < 2.2 × 10−16], and functional versus random dbSNPs [HEK293: p = 2.72 ×10−09, HeLa: p = 9.74 × 10−09]). d Relative activity of MapUTR variants is corroborated by predicted changes in RBP binding. X-axis similar as (c). Functional variants (blue) that increase ZFP36 binding to AU-rich elements WUUAUU and AUUUAU in HEK293 and HeLa, respectively, exhibit decreased gene expression. Nonfunctional variants (red) cause little changes in gene expression regardless of changes in RBP binding. The shaded bands indicate the 95% confidence interval from the line of best fit. See also Supplementary Fig. 10b. a, b All boxplots depict the median as the center line, the boxes define the interquartile range (IQR: 25th to 75th percentiles) and the whiskers extend up to 1.5 times the IQR. a–d Source data are provided as a Source Data file.
