Fig. 2: Community detection algorithm identifies hiPSC regulatory networks. | Nature Communications

Fig. 2: Community detection algorithm identifies hiPSC regulatory networks.

From: Complex regulatory networks influence pluripotent cell state transitions in human iPSCs

Fig. 2

A Diagram of proposed molecular mechanisms underlying gene co-expression networks. Co-expression of genes located on different chromosomes but within the same GNM (A or B) is mediated via regulatory elements that are co-accessible because they bind the same transcription factors (TF1 or TF2). Created with BioRender.com. B Histogram displaying the number of peaks in each of the 13 major RNMs. C UMAP plot of 4772 ATAC-seq peaks in the 13 major RNMs. We plotted ATAC-seq peaks with the top 10% intramodular co-accessibility in each RNM, as opposed to the Pareto peaks (n = 9545) for plot legibility. Each point represents a peak colored by its corresponding RNM. Source data are provided as a Source Data file. D Heatmap showing the pairwise associations between 13 RNMs based on the co-accessibility enrichment. Each cell is filled with the log2(Odds Ratio) for the corresponding RNM pair. Source data are provided as a Source Data file. E Barplot showing the estimated cell state proportions across 150 ATAC-seq samples. Previously published ATAC-seq peaks from FACs sorted cells representing formative and primed pluripotency states1 were used to perform cellular deconvolution on the 150 iPSCORE hiPSC ATAC-seq samples. Each stacked barplot corresponds to an ATAC-seq sample and the colors correspond to the estimated formative and primed cell states. Source data are provided as a Source Data file. FH Boxplots demonstrating RNM 2 (F), 3 (G), and 8 (H) associations with the estimated proportion of cells in the formative state. A linear model was used to calculate the association between RNM scores (summed inversed normal transformed accessibility of RNM Pareto peaks) and the estimated formative proportion of 123 ATAC-seq samples (<100% estimated formative high proportion). Each point represents an ATAC-seq sample. Samples were binned by estimated proportion for boxplot legibility. The boxplot maxima and minima are set by the samples with the highest and lowest RNM score, the center line represents the RNM score, the upper and lower hinges correspond to the 25th and 75th percentiles, and the upper and lower whiskers extend to the highest and lowest value within 1.5 times the interquartile range (IQR). Source data are provided as a Source Data file. I Heatmap showing GNM-RNM associations. 32,327 ATAC-seq peaks in the 13 major RNMs were annotated with a putative target gene (see “Methods”). We performed pairwise Fisher’s Exact test to calculate enrichments (log2(Odds Ratio)) of RNM peaks in GNMs. Non-significant associations are filled in white. Source data are provided as a Source Data file.

Back to article page