Supplementary Figure 1: Evaluation of allelically informative accessible regions and monoallelic sites.

(a) Comparison of allelic read assignment for technical replicates, biological replicates and different clones. Pearson’s correlation coefficients were calculated from allelic counts for the 129 and Cast alleles. (b) Comparison of two statistical methods for evaluating allelic bias based on the binomial distribution and permutation method. Cumulative curves for P value (adjusted by the Benjamini–Hochberg procedure) from the X chromosome and chromosome 5 are shown in red and blue, respectively. The permutation method was compared to the exact binomial test. The right panel is a zoomed-out view of the left panel. The difference between the dark and light curves shows that the permutation method is more stringent on the autosomes, but more sensitive on the X chromosome. (c) Scatterplot showing the number of usable reads versus the number of allelically informative accessible regions. The number of allelically informative regions is close to stable when the number of usable sequencing reads is more than 40 million. The number of allelically informative accessible regions in ESCs is lower than for NPC clones with the same number of usable reads (red points). Blue points show downsampled data from a highly sequenced line (NPC XY14). (d) Scatterplot of TSS enrichment score versus the number of allelically informative accessible regions. There is no significant correlation between them (r = –0.186, P = 0.461). Red indicates the two ESC lines. (e) Simulation to determine the number of reads needed to identify monoallelic peaks in NPCs. 6–100 million usable reads were randomly sampled from a deeply sequenced NPC clone (NPC XY14). The number of monoallelically accessible regions dramatically increases up to 40 million reads where it plateaus. (f) Number of monoallelic peaks identified by ATAC–seq on the Cast X chromosome, 129 X chromosome, Cast autosomes and 129 autosomes in each cell line analyzed.