Extended Data Fig. 3: Cis-regulatory atlas in rice at single-cell resolution.

a, Bar plot showing the proportion of number of ATAC peaks in different genomic regions of scATAC in different organs. Union peaks refer to the peaks that are located relatively far from the TSS (>3 kb). b, scATAC fragment length distributions for each organ. c, Enrichment of scATAC peaks surrounding TSSs for each organ. d, Ridge plot showing TSS enrichment score for each organ. e, Distribution of different epigenetic markers about middle point of ACRs. f, Heatmap showing scATAC and scRNA correlation (Spearman correlation coefficient) between cell types, respectively. The scATAC correlation was calculated based on the average peak counts per cluster, while the scRNA correlation was calculated based on the average gene expression per cluster. g, Cluster-aggregated chromatin accessibility (left) and RNA expression (right) are shown at the DNA BINDING WITH ONE FINGER6 (Dof6) locus. h, Identified seven regulon modules based on regulon connection specificity index (CSI) matrix, along with representative transcription factors and corresponding binding motifs. The rice RcisTarget database was utilized to identify TF binding motifs and predict target genes. In the heatmap, the colour intensity corresponds to the Regulon Activity Score (RAS), with darker colours indicating higher regulatory activity. i, A zoomed-in view of module M7 and its sub-modules is presented. j, Sankey plot showing correlation between cell-type and regulon modules M1-M7. The regulon module activity score was computed for each cell, and mean values for different cell types were visualized using a Sankey diagram, created with the ‘sankeyNetwork’ function from the ‘networkD3’ package (v.0.2.4) in R. The thickness of the connected lines represents the strength of correlation, and the 7 coloured lines represent the strongest correlations for each module. k, Heatmap showing the expression patterns of sub-modules of M7 across different cell types. Correlation coefficients between regulons were derived from RAS values to calculate the Connection Specificity Index (CSI). Regulons were hierarchically clustered based on CSI to define regulon modules, and a heatmap was produced using the ‘ComplexHeatmap’ package (v.1.99.4).