Fig. 1: An overview of the scMTNI framework.

a scMTNI takes as input a cell lineage tree and cell type-specific scRNA-seq data and cell type-specific prior networks derived from scATAC-seq datasets. If scATAC-seq data is not available, bulk or sequence-based prior networks can be used for the cell types. The output of scMTNI is a set of cell type-specific gene regulatory networks for each cell type on the cell lineage tree. b The output networks of scMTNI are analyzed using two dynamic network analysis methods: edge-based k-means clustering and Latent Dirichlet Allocation (LDA) based topic models to identify key regulators and subnetworks associated with a particular cell cluster or a set of clusters on a branch. c Datasets used with scMTNI. The simulation data comprised a linear trajectory of three cell types, while the three real datasets came from a reprogramming time-series process, immunophenotypic cell types identified during human adult hematopoietic differentiation, and immunophenotypic blood cells during human fetal hematopoiesis. MEF mouse embryonic fibroblast, iPSCs induced pluripotent cells, HSC hematopoietic stem cell, CMP common myeloid progenitor, GMP granulocyte-macrophage progenitors, Mono monocyte, HSC-MPP hematopoietic stem cells and multipotent progenitors, LMP lymphoid-myeloid progenitors, MEMP MK-erythroid-mast progenitors combined with cycling MEMPs, GP granulocytic progenitors, Ery erythroid cells, pDC plasmacytoid dendritic cells.