Fig. 4

Integrated analysis of PBMC scRNA-seq and scATAC-seq data using MAESTRO. a UMAP visualization for joint clustering of human PBMC scRNA-seq (12k cells) and PBMC scATAC-seq (10k cells). Colors represent the cells from different technologies. The cells are joined by CCA on gene expression level and regulatory potential from MAESTRO. b UMAP visualization for joint clustering of human PBMC scRNA-seq and scATAC-seq. The cells are joined by CCA on the gene expression level and regulatory potential from MAESTRO. Colors represent the cell types, for which are generated using the scRNA-seq dataset and transferred to the scATAC-seq dataset. c The rank of driver regulators in CD14 monocyte cells of the PBMC dataset. The x-axis represents the TF enrichment score from LISA results in cluster-specific genes using scRNA-seq; the y-axis represents the TF enrichment score from GIGGLE results in cluster-specific peaks using scATAC-seq. The color of the circles represents the averaged expression level of corresponding regulators in CD14 monocyte scRNA-seq cells, and the size represents the TF enrichment score using GIGGLE in CD14 monocyte scATAC-seq cells. The names of the top 10 TFs from LISA and GIGGLE are labeled on the graph. d Comparison of transcriptional regulators predicted using scRNA-seq and scATAC-seq in each cell type for PBMC dataset. The y-axis represents the Spearman’s correlation coefficient between LISA-predicted TF enrichment score and GIGGLE-predicted TF enrichment score for all the tested regulators. e Genome browser view of MS4A1 (B cells), CD8A (T cells), and HLA-DQA1 (monocytes and DCs) locus. The pseudo-bulk ATAC-seq profiles are generated by pooling together cells within each cell type. The y-axis represents the sequence depth-normalized ATAC-seq signals (reads per million mapped reads (RPM))