Fig. 7
From: Genome-wide prediction of DNase I hypersensitivity using gene expression
Predicting differential activities of regulatory elements during neuronal differentiation. a Difference in H3K4me1 measured by ChIP-qPCR between iPSCs and iPSC-derived neurons for 26 DHSs at different predicted fold change levels (i.e., absolute value of log2 fold change \(\left| \delta \right| >2\), 1 < \(\left| \delta \right|\) ≤ 2, and 0.1 < \(\left| \delta \right|\) ≤ 1, indicated by yellow, green, and blue, respectively, in the gene labels) and five non-differential DHSs (“control”) predicted by BIRD. The 26 DHSs were further divided into up regulated and downregulated DHSs after differentiation based on the predicted DH. Dashed lines represent ± maximum |ChIP-qPCR measured H3K4me1 difference in five control DHSs|. Data are presented as mean ± s.e.m. (iPSCs and iPSC-derived neurons each had n = 3 technical replicates). b Validation rate of differential DHSs: 7/12 high-ranked (\(\left| \delta \right| >2\)) DHSs, 8/10 middle-ranked (1 < \(\left| \delta \right|\) ≤ 2) DHSs, and 1/4 low-ranked (0.1 < \(\left| \delta \right|\) ≤ 1) DHSs were validated by ChIP-qPCR. c Correlation between the predicted differential DH (i.e., difference in the predicted DH signals (at log2 scale) between iPSCs and iPSC-derived neurons) and the ChIP-qPCR measured H3K4me1 difference (log2 fold change) across the 31 DHSs tested by ChIP-qPCR. Each dot is a DHS. The Pearson’s correlation coefficient is shown on the top
