Fig. 6
From: Single-cell trajectories reconstruction, exploration and mapping of omics data with STREAM
STREAM on single-cell epigenomic data from the human hematopoietic system. a Single-cell ATAC-seq workflow. FACS sorting is used to isolate populations from CD34+ human bone marrow and single-cell ATAC-seq measurements are performed. After mapping reads to the reference genome, reads within peaks are selected and ChromVAR is used to calculate k-mers z-scores. Finally, PCA is applied to the z-score matrix and top principal components are selected as features for the STREAM analysis. b STREAM learns a principal graph from chromatin-accessibility data and accurately reconstructs cellular developmental trajectories of the human hematopoiesis. As in Fig. 1, the structure can be easily visualized thanks to the subway map and stream plots. In the first branch, the HSCs segregate through MPP into lymphocyte-committed, erythrocyte-committed and myelocyte-committed branches. STREAM also reconstructs the bifurcation from lymphoid multipotent progenitors (LMPP) to CLP and plasmacytoid dendritic cells (pDC). c Discovery of transcription factors important for lineage commitment. 7-mer DNA sequences are automatically detected and their frequencies are visualized in both the subway map and stream plots. These sequences are mapped to known transcription factors motifs. We recovered GATA1 and CEBPA as top hits, two classic master regulators in blood development, which correlate with directionality toward erythroid differentiation and myeloid differentiation, respectively
