Fig. 3: Chromatin accessibility reveals alteration of transcriptional networks in MDS.

a Heat map showing the enrichment of transcription factor binding motifs at the top 15% ATAC peaks of each disease stem and progenitor fraction. Z-scores of -logP values are shown as motif enrichment scores. Transcription factor-binding motifs whose -logP value of motif enrichment was higher than 30 at the DARs of any disease subtype, as shown in Supplementary Fig. 6, were selected. The HSPC fractions in which the binding motif of each transcription factor was most enriched in Fig. 2f are indicated at the bottom. b Pearson correlation coefficients between motif enrichment scores and gene expression of the corresponding transcription factors in MDS stem (left) and progenitor (right) samples. Motif enrichment of the transcription factor-binding motifs was calculated by averaging the ATAC peak values of MDS DARs (DARs between MDS-MLD or MDS-EB and normal samples) with the corresponding binding motifs. c Correlation between the enrichment of transcription factor-binding motifs at DARs and the expression of corresponding transcription factor genes during disease progression. d Transcription factor networks of MDS stem (left) and progenitor cells (right) constructed by the GeneNet method of GeNeCK using the motif enrichment score of each sample indicated in Fig. 3b and visualized using Cytoscape. The color and width of the edge indicate the partial correlation coefficient, reflecting the strength of the linear relationship between two transcription factor motifs, independent of other motifs. The size of the oval node represents the maximum enrichment score in the disease group. As the AP-1 family motifs showed high enrichment scores at closed DARs, they are depicted as relatively large blue ovals. Motifs were selected based on the criteria described in Fig. 3a for MDS stem and progenitor cells. Source data are provided as a Source Data file.