Fig. 5

Detection of regulated sequences in ΔBCL7B cells identified by machine learning. a The ChIP-seq data analysis workflow. b-d From ChIP-seq data of the control parent cell line and ΔBCL7B cell lines, 372,870 regions (Peaks2.0) and 367,631 regions (Peaks0.5) were identified as peaks preferentially obtained in the ΔBCL7B group and control group, respectively. b All the converted peak information is displayed as scatter plots after UMAP embedding. The colour coding is based on the clusters (left panel) or the ratios of read depth of the ΔBCL7B group to the control group (right panel). c Cluster centroids clusters are shown in a line chart (upper left pane). The horizontal axis indicates the index of short motifs, and the vertical axis indicates the frequency of each motif. The centroids of Clusters 3 and 4, that consisted only of peaks derived from the ΔBCL7B group, indicate high frequency of short motifs #13 and #49, respectively. The Clusters 5 and 7 preferentially consisted of peaks derived from the control group. The centroids indicate high frequencies of short motifs #79 and #60. d The sequence logos of short motifs #13 and #49 (left panels) and #79 and #60 (right panels) are shown. e The RNA-seq data indicated that the genes downregulated in the ΔBCL7B group and located within 10 kb of the peaks belonged to Clusters 3 and 4 and are shown in tables. Immune response-related genes (blue) and the MIRLET7A gene (purple) were identified