Fig. 1: Activity of regulatory variants linked to polygenic skin disease.

a Epidermal schematic. b Hypothetical model for one mechanism whereby non-coding variation may affect polygenic skin disease risk. c MPRA experimental diagram for assaying variants correlated with skin disease (day 0, undifferentiated keratinocyte populations, and day 3 and 6 of calcium induced differentiation). d PCA of MPRA activity (linear scale, outliers clipped) across differentiation shows distinct transcriptional states. e PCA of transcriptional activity in keratinocytes treated with CRISPR guides knocking out IRF6 or SNAI2 TFs rotate cells through the transcriptional states of differentiation. “Safe” cells were treated with a safe targeting guide. f Activities of known skin TFs correlate with transcriptional activity. Effect size (Hedge’s g, or the difference in means divided by the pooled standard deviation, with a correction factor for small samples) measures the difference in element activity for fragments with the given motif vs fragments without, and the P value is the two-sided Mann–Whitney test for a difference in activity. When relative to day 0, the day 0 activity is first subtracted from each element. g Association of transcriptional activity with JASPAR motif clusters (as fold increase in activity for elements with the motif) suggests SP/KLF motifs are the strongest in promoting transcription, followed by AP-1 and NRF1 motifs. h Effect sizes and significance values (hue, two-sided Mann-Whitney) for a change in motif-activity associations across differentiation; note transiently increased activity of SP/KLF, NRF1, and AP-2 motifs at day 3, and sustained activity of CEBP/ATF and AP-1 motifs from early to late differentiation. Gmeb1 is the glucocorticoid receptor binding site, which is transiently activated, then repressed. P value, Mann–Whitney. i daSNVs with dynamic activity (FDR < 0.001, top 20 by fold change) across differentiation increase activity and reside near differentiation genes.