Fig. 5: Gained and lost activity eSNPs disrupt binding of FOX TFs and are associated with complementary mechanisms.

Effect size (y-axis) of gained (A) and lost (B) activity eSNPs associated with enriched TFs against their putative target gene counts (x-axis). Only the TFs with a number of gene targets ≥15 and an average EWN≥7 are labeled in the figure. C Fold enrichment of different cancer signatures in FOXA1-associated gained activity eSNPs compared to FOXA1-associated lost activity eSNPs. D FOXA1 binding signal change (MDA - LNCaP) at gained (n = 56) and lost (n = 62) activity eSNP positions where MDA PCa 2B (dubbed MDA) genotype has a greater number of alternate alleles. The non-eSNP positions (n = 7933) are used as background. Bonferroni-corrected p-values are obtained from a one-sided Wilcoxon test. In the boxplots, the horizontal line in the middle is the median value, and the lower and upper edges of the boxes correspond to the 25th and 75th percentiles. Extending vertically upwards/downwards of the boxes are the lines showing 1.5 times the interquartile range (i.e., distance between 25th and 75th percentile). E Read coverage of FOXA1 at rs10095018 in LNCaP and MDA cell lines, where the alternate allele C disrupts the binding of FOXA1 according to the consensus motif of FOXA1 (see Fig. S5B for larger motif logo). The graphic of the genomic coordinates with the track of FOXA1 reads coverage was generated using UCSC genome browser⁶⁹ (http://genome.ucsc.edu/; UCSC hg19 assembly). F Gene expression measured as log₂(TPM + 1) of NDRG1, a likely target of an enhancer harboring rs10095018, in two prostate cancer cell lines: MDA and LNCaP. Source data for these figures is provided as a Source data file.