Fig. 1

Inference of cell-type-specific gene networks from single-cell transcriptome data and their applications in dissecting disease genetics. A scHumanNet framework that infers cell-type-specific networks (CGNs) through filtering reference interactome, HumanNet, and analytical modules for deriving differential hub genes and deconvoluting gene sets based on network connectivity are described. B De novo inference of CGNs through various approaches of preprocessing single-cell transcriptome data. The three methods with independent Pearson correlation coefficients (PCC) scores as gene–gene edge weights were converted to log-likelihood score (LLS). High LLS score edges are kept and integrated using the identical approach modeling HumanNet, integrating edges from various sources of evidence. C A compendium of reference CGNs inferred from human cell atlas data, demonstrating their utility in identifying disease-associated genes by comparing them with CGNs inferred from disease samples of the same organ