Differential gene network analysis from single cell RNA-seq.

Study of gene expression has been arguably the most active research field in functional genomics. Over the last two decades, various high-throughput technologies, from gene expression microarray to RNA-seq, have been widely applied to the wholegenome profiling of gene expression. The commonality of these experiments is that they measure the gene expression levels of “bulk” sample, which pools a large number (often in the scale of millions) of cells, and thus the measurements reflect the average expression of a population of cells. The recently developed single-cell RNA sequencing technology (scRNA-seq) allows the transcriptomic profiling at the single-cell level (Tang et al., 2009). Compared with the bulk experiments, scRNA-seq provides important information for inter-cellular transcriptomic heterogeneity, adding another dimension to understand gene expression regulation and dynamics. The technology has gained considerable interests recently, and a number of experiments have been performed to study highly heterogeneous samples such as cancer and brain (Patel et al., 2014; Zeisel et al., 2015). One major goal of scRNA-seq experiment is to characterize the heterogeneity of gene expression among cells, and then relate that to phenotypic variation. So far, the scRNA-seq data analyses have beenmainly focused on cell clustering (Bendall et al., 2014; Trapnell et al., 2014) and differential expression (Kharchenko et al., 2014). However, since genes function through a complex biological system, another important aspect of gene expression analysis is to reconstruct and detect changes in the gene networks (Hase et al., 2013; Siegenthaler and Gunawan, 2014). Differential network analysis can reveal biological responses to stimuli through the rewiring of biological network (Ideker and Krogan, 2012). So far, the network analysis of scRNA-seq data has not been fully explored.