O05_05

SGCRNA: A Novel Tool for Gene Co-Expression Network Analysis Using Spectral Clustering

Tatsunori OSONE *, Takeshi TAKARADA

Department of Regenerative Science, Okayama University
( * E-mail: osone@okayama-u.ac.jp)

Weighted Gene Co-Expression Network Analysis (WGCNA) is a potent methodology that is capable of identifying functional modules and pivotal genes through the analysis of gene co-expression patterns. The utilisation of WGCNA for the structural analysis of gene expression data enhances the comprehension of intricate biological processes, thereby leading to its widespread adoption in various studies. Since its inception in 2005, numerous advanced tools, including applications to single-cell RNA-seq, have been developed. However, the scale-free nature of networks, a premise of WGCNA, was widely accepted two decades ago when WGCNA was first proposed. Recently, it has been reconsidered that only a limited number of networks exhibit these properties. Additionally, from a usability perspective, WGCNA necessitates manual parameter adjustments in at least two steps. Biologically, although correlations are considered, the ratios between genes are not, and negative correlations are fundamentally disregarded. To address these four points, a novel method has been developed. By applying this method to publicly available data from bulk RNA-seq, single-cell RNA-seq, and spatial transcriptome analyses, biologically meaningful results have been obtained in all cases. Compared to the outcomes derived from conventional methods, the new method has demonstrated higher utility and increased depth of analysis. These findings suggest that the new method can produce biologically significant results within a reasonable execution time without relying on arbitrary parameter adjustments. As this study lacks wet lab validation, it is necessary to empirically verify whether the hub genes and their putative controlling transcription factors predicted by the new method are accurate.