**1. Introduction**

*Dendrobium* is a perennial epiphytic herb of the genus *Orchidaceae*, with more than 1500 species in the world which are mostly growing in tropical and subtropical Asia and eastern Australia [1,2]. In China, there are more than 80 species of *Dendrobium* that had been reported in studies, mainly distributed in south of the Qinling Mountains [3,4]. As a kind of ornamental and medicinal plants, more and more researchers have paid attention to it in recent years [5]. Also, *Dendrobium* is a traditional Chinese herbal medicine, that has been used for benefiting stomach and clearing heat, nourishing yin, and promoting fluid [6]. The medicinal ingredients of *Dendrobium* are very complex, not only including polysaccharides and alkaloids, but also flavonoids, various amino acids, and trace mineral elements. However, the most two important components are polysaccharides and alkaloids [7–10].

With the development of high-throughput technology, a variety of data sources have been derived, including gene expression microarray, RNA-seq, metabolomic and CHIP-seq, these have become powerful tools for studying plant growth, development, and physiology at the transcriptional and metabolic levels [11–13]. In order to explore the key genes and regulatory pathways for the synthesis of its medicinal ingredients, more researchers have done different kinds of *Dendrobium* transcriptome

research [14–16]. However, it has been difficult to systematically explain the relationship between gene expression or metabolite changes and trait differences [17]. The analysis of correlation networks can bridge the gap between single gene interpretation and systematic biology research by mining the link between genes and gene products [18–20], such as integrating a single gene into a co-expression network based on pairwise gene expression correlation [21]. Gene co-expression analysis has been used to discover new candidate genes [22,23], identify key modulators of immune responses [24], and reconstruct regulatory pathways [25]. Genes belonging to the same co-expression sub-network (or module) are likely to be functionally related [26–28], participate in similar biological processes, or be part of the same pathway [25].

Weighted gene co-expression network analysis (WGCNA) is one of the most useful methods based on gene co-expression networks. It focuses on the set of genes other than on a single gene in the observed gene expression data, and it alleviates the multiple detection problems inherent in chip data analysis and can be used in unweighted correlation networks [29]. Compared with many other analysis methods, WGCNA has the advantages of summarizing and standardizing the methods and functions of integrated R packages, including methods of weighted and unweighted correlation networks [30]. WGCNA is used in combination with gene chip data, transcriptome data, and metabolome data for metabolic regulation network simulation, mining inter-genetic interactions, screening functional genes, etc. It has been extensively studied in plant growth, tissue and organ development, pigmentation and fragrance synthesis [31–34].

To reveal the underlying molecular mechanism of the active ingredients of *Dendrobium*, we downloaded three datasets from the NCBI Sequence Reading Archive (SRA) to identify highly connected hub genes and important modules. This study used three different *Dendrobium* species and different tissues as materials to perform transcriptome sequencing data, combined with polysaccharide and alkaloid content data, and used WGCNA analysis to construct a co-expression gene network. Correlation analysis was performed between the gene module and the polysaccharide and alkaloid data, and hub genes related to the main medicinal ingredients were discovered, in order to provide new clues for further research on the molecular mechanism of medicinal ingredients of *Dendrobium*. For the first time, a co-expression network analysis of transcriptome genes in *Dendrobium* was constructed, and modules with high correlation in secondary metabolism were analyzed, laying a foundation for the discovery of functional genes of medicinal ingredients.
