# Comparing gene-gene co-expression network approaches for the analysis of cell differentiation and specification on scRNAseq data

## TL;DR

This paper compares different methods for analyzing gene networks in single-cell RNA data to understand cell differentiation and development over time.

## Contribution

The study reveals that analysis strategy, not network model choice, most strongly affects results in gene co-expression network analysis of cell differentiation.

## Key findings

- Network analysis strategy has a stronger impact on results than network modeling choice.
- Combined time point modeling is more stable than single time point modeling.
- Differential gene expression-based methods best model cell differentiation.

## Abstract

Gene-gene co-expression network analysis has been widely applied to bulk RNA sequencing and microarray data to investigate different phenotypes and compound exposures. Recently, it has also been applied to single cell RNA sequencing data. However, the impact of different network models, data processing pipelines, and analysis strategies on downstream interpretations has not yet been characterized.

Here we study the impact of network models and analysis strategies on the resulting interpretations from analyses of cell differentiation and cell state over time using gene-gene co-expression networks.

Our results suggest that the network modeling choice has less impact on downstream results than the network analysis strategy selected. The largest differences in biological interpretation were observed between the node-based and community-based network analysis methods (strategies). In addition, we observe a difference between single time point and combined time point modeling.

•We investigated gene-gene co-expression network modeling approaches.•Combined time point modeling performed more stable than single time point modeling.•Differential gene expression-based methods model cell differentiation the best.•Network analysis strategy has the strongest impact on the results.

We investigated gene-gene co-expression network modeling approaches.

Combined time point modeling performed more stable than single time point modeling.

Differential gene expression-based methods model cell differentiation the best.

Network analysis strategy has the strongest impact on the results.