Eucalyptus grandis Auxin Response Factor 10 (EgrARF10) is associated with the modulation of secondary xylem cell wall chemistry
Data files
Jul 08, 2026 version files 2.56 GB
-
euc_multinetwork.Rdata
2.56 GB
-
README.md
5.32 KB
Abstract
Auxin Response Factors (ARFs) are of vital importance in plant growth and vascular development. Class A ARFs are the core auxin-driven drivers of embryogenesis, vascular pattern development, (pro)cambial stem cell initiation and xylem cell fate specification, while class B ARFs may attenuate or fine-tune auxin-mediated development. The potential roles of class C ARFs in vascular development and xylogenesis, however, remain largely unexplored. In this study, we identified Eucalyptus grandis ARF10 (EgrARF10) as a potential regulator of secondary cell wall (SCW) development. A supra-adjacency matrix consisting of seven gene-gene networks (“layers”) was constructed with RWRToolkit (Kainer et al., 2025), with equal weighting between each layer. The layers included multi-omics networks built from published co-expression data in E. grandis , trans-eQTL data, A. thaliana and P. trichocarpa predicted expression networks built from iterative Random Forest Leave-One-Out Prediction algorithms, DAP-seq data, protein-protein interaction data, and a curated Arabidopsis secondary cell wall transcriptional regulatory network. Where Arabidopsis and Populus gene-gene networks were used, the Integrative Orthology tool in PLAZA Dicots 5.0 (Van Bel et al. 2022) was used to identify tree-based orthologs in E. grandis. Then, the Random Walk with Restart – Lines Of Evidence (RWR-LOE) function in RWRToolkit was run with EgrARF10 as seed gene to rank the proximity of all genes in the supra-adjacency network relative to EgrARF10.
Dataset DOI: 10.5061/dryad.q573n5tzv
File: euc_multinetwork.Rdata
Description: 7-layer multinetwork constructed with the RWR_make_multiplex() function in RWRToolkit.
Description of the data and file structure
File: euc_multinetwork.Rdata
This file is an R data object file (.Rdata) containing a seven-layer multiplex gene network for Eucalyptus grandis. The network was constructed using RWR_make_multiplex from RWRtoolkit, with equal weighting between each layer, and is intended for downstream random walk with restart analyses, including network proximity ranking, gene-set prioritisation and multi-omic network exploration.
The multiplex network integrates seven biological evidence layers derived from publicly available Arabidopsis thaliana, Eucalyptus grandis and Populus trichocarpa datasets, together with network layers generated or lifted over as described in the associated manuscript. The seven layers are stored in the multiplex object as:
coex(a Pearson's co-expression layer, containing 15,606,296 interactions, from 479 Eucalyptus RNA-seq datasets)Ath_PEN(an A. thaliana Predictive Expression Network using iterative Random Forest-Leave One Out Prediction, containing 9,429,983 interactions and assigning E. grandis orthologs based on best BLAST hit)Pop_PEN(a P. trichocarpa Predictive Expression Network using iterative Random Forest-Leave One Out Prediction, containing 3,129,044 interactions and assigning E. grandis orthologs based on best BLAST hit)eQTL_ur_F1hyb(a network of 1,263,357 E. grandis × E. urophylla backcross trans-eQTLs involving transcription factors, with edges weighted by co-expression correlation across the population)DAP-seq(a network of 1,225,116 predicted TF-target interactions from DNA Affinity Purification Sequencing data in Eucalyptus and Arabidopsis)PPI_liftover(an A. thaliana protein-protein interactions network of 63,102 interactions, assigning E. grandis orthologs based on best BLAST hit)SCW_network_liftover(a set of 814 curated secondary cell wall-related TF-target gene interactions from A. thaliana, assigning E. grandis orthologs based on best BLAST hit)
These layers represent distinct forms of gene–gene relationship evidence, including Eucalyptus co-expression data, trans-eQTL data from Eucalyptus, predicted expression networks from Arabidopsis and Populus, DAP-seq-derived regulatory relationships, protein–protein interaction data, and a curated secondary cell wall transcriptional regulatory network.
Loading the file in R creates three objects:
load("euc_multinetwork.Rdata")
ls()
# "nw.adj" "nw.adjnorm" "nw.mpo"
Object descriptions
nw.mpo is the main RWRtoolkit Multiplex object. It contains the seven named network layers listed above, together with the pool of nodes, the number of nodes in the multiplex and the number of layers.
nw.adj is the sparse supra-adjacency matrix for the multiplex network, stored as a dgCMatrix object from the R Matrix package. Its dimensions are 209,993 × 209,993. Rows and columns correspond to gene–layer combinations rather than genes alone. Because the network contains seven layers, this corresponds to 29,999 gene nodes represented across seven network layers.
nw.adjnorm is the normalized sparse supra-adjacency matrix, also stored as a dgCMatrix object with dimensions 209,993 × 209,993. This matrix is used by RWRtoolkit for random walk with restart analysis across the multiplex network.
Supra-adjacency matrix structure
The supra-adjacency matrix preserves the multilayer structure of the network. Each diagonal block represents gene–gene edges within one evidence layer. Off-diagonal blocks represent interlayer links connecting the same gene across different evidence layers. This structure allows random walks to move both within individual biological evidence layers and between layers, while retaining the distinct topology of each network layer rather than collapsing all evidence into a single aggregated network.
Software requirements
The file is intended for use in R with RWRtoolkit and its dependencies. Loading the object requires the R Matrix package because the supra-adjacency matrices are stored as sparse dgCMatrix objects. Additional downstream analyses require RWRtoolkit and associated Bioconductor/R dependencies used by that package.
Intended reuse
This file can be reused to perform RWRtoolkit-based multi-omic network analyses in Eucalyptus grandis. It is suitable for ranking genes by network proximity to user-defined seed genes, identifying genes connected across multiple lines of biological evidence, evaluating gene-set connectivity and exploring secondary cell wall-associated regulatory modules.
Code/software
Access information
Other publicly accessible locations of the data:
-
None