Application of an improved correlation network analysis tool, HIGOMARI, for gene co-expression analysis in plants

Gene co-expression analysis is a data analysis technique used to identify gene modules, defined as groups of genes that exhibit similar expression patterns under various experimental conditions. Based on ConfeitoGUI, a correlation network analysis tool, we developed the HIGOMARI tool to better explore relationships between modules by incorporating a topological modularization framework that enables explicit visualization of network topology. Unlike conventional methods that rely primarily on clustering approaches, HIGOMARI refines module structures through false-positive-out (FPO) and false-negative-in (FNI) analyses. These processes enable the removal of spurious associations while recovering biologically relevant but weak connections, thereby improving module reliability and interpretability. To evaluate the effectiveness of HIGOMARI, we performed two gene co-expression analyses. First, we analyzed large-scale transcriptome data from Arabidopsis thaliana, focusing on gene modules involved in secondary metabolism regulated by MYB transcription factors, including the flavonoid and glucosinolate biosynthesis pathways. HIGOMARI efficiently identified these modules and enabled the analysis of relationships between adjacent modules, revealing connections that are difficult to detect using conventional approaches. Second, we re-analyzed gene modules associated with the triterpenoid biosynthesis pathway of cucurbitacin, a bitter compound in bitter melon (Momordica charantia), using RNA-sequencing data. In addition to the three previously identified P450 genes, HIGOMARI identified a module containing one additional uncharacterized P450 gene, two glycosyltransferase genes, and two transcription factor genes potentially involved in cucurbitacin biosynthesis. These results demonstrate that HIGOMARI provides a flexible and scalable framework for simultaneously exploring intra-module structure and inter-module relationships, offering a powerful extension to existing gene co-expression analysis methodologies.