Constructing local cell-specific networks from single-cell data

Abstract

Gene coexpression networks yield critical insights into biological processes, and single-cell RNA sequencing provides an opportunity to target inquiries at the cellular level. However, due to the sparsity and heterogeneity of transcript counts, it is challenging to construct accurate gene networks. We develop an approach, locCSN, that estimates cell-specific networks (CSNs) for each cell, preserving information about cellular heterogeneity that is lost with other approaches. LocCSN is based on a nonparametric investigation of the joint distribution of gene expression; hence it can readily detect nonlinear correlations, and it is more robust to distributional challenges. Although individual CSNs are estimated with considerable noise, average CSNs provide stable estimates of networks, which reveal gene communities better than traditional measures. Additionally, we propose downstream analysis methods using CSNs to utilize more fully the information contained within them. Repeated estimates of gene networks facilitate testing for differences in network structure between cell groups. Notably, with this approach, we can identify differential network genes, which typically do not differ in gene expression, but do differ in terms of the coexpression networks. These genes might help explain the etiology of disease. Finally, to further our understanding of autism spectrum disorder, we examine the evolution of gene networks in fetal brain cells and compare the CSNs of cells sampled from case and control subjects to reveal intriguing patterns in gene coexpression.

Cell-type-specific co-expression inference from single cell RNA-sequencing data

“…We compared CS-CORE to other approaches, including locCSN (Wang et al, 2021b), Noise Regularization (Zhang et al, 2021), Normalisr (Wang, 2021), Pearson correlation, propr (Quinn et al, 2017), ρ -sctransform (Hafemeister and Satija, 2019), Spearman correlation and SpQN (Wang et al, 2022) (Section 4.2). Among these approaches, statistical tests for co-expressions are possible for Noise Regularization, Normalisr, Pearson correlation, ρ -sctransform and Spearman correlation.…”

“…We compared CS-CORE with eight other methods for inferring gene co-expression from single cell data, including locCSN (Wang et al, 2021b), Noise Regularization (Zhang et al, 2021), Normalisr (Wang, 2021), Pearson correlation, propr (Quinn et al, 2017), ρ -sctransform, Spearman correlation and SpQN (Wang et al, 2022). The method locCSN was applied on log normalized data log( x ij /s i + 1) and computed with the Python implementation provided at https://github.com/xuranw/locCSN.…”

“…Recent years have seen the developments of a number of methods to better capture coexpressions from scRNA-seq data than a simple normalization-based approach, including locCSN (Wang et al, 2021b), Noise Regularization (Zhang et al, 2021), Normalisr (Wang, 2021), propr (Quinn et al, 2017), and SpQN (Wang et al, 2022). These methods consider novel association metrics or additional adjustments when inferring co-expressions from scRNA-seq data.…”

spliceJAC : transition genes and state‐specific gene regulation from single‐cell transcriptome data

“…Recently, other interesting methods started to focus on the cell state specificity of gene regulation. The cell type‐specific gene networks in the developing fetal brain and in autistic patient samples has been developed, relying on a local independent test, which is carried out on a cell type‐basis (Wang et al , 2021); GRNs of distinct cell types have been constructed by integrating prior knowledge and gene activity (Gibbs et al , 2022); a fine‐grained method has been developed to infer cell‐specific networks (CSN) and predict important genes that are neglected by traditional differential gene expression analysis (Dai et al , 2019). One common trait of these methods is to adjust the global GRN inference strategies, such as correlation, mutual information, and regression, and apply them to specific cells or subpopulations of cells (Akers & Murali, 2021).…”

RNA degradation heavily impacts mRNA co-expression

“…On the other hand, gene co-expression in scRNA-seq datasets has been exploited to extrapolate the functional principles of gene expression regulation (Desai et al, 2021) from cellular populations in the form of gene-regulatory networks (Matsumoto et al, 2017). The most common approach is the construction and analysis of co-expression networks from pairwise correlation measurements (Vivian Li and Li, 2021; Wang et al, 2021). Single-cell co-expression network analysis has thus led to the characterization of novel regulatory pathways (Xie et al, 2021), which in turn has improved our understanding of common gene regulation principles between cell types and species (Crow et al, 2022; Harris et al, 2021).…”