Divergence of Root Microbiota in Different Habitats based on Weighted Correlation Networks
September 25th, 2021
Network analysis was applied to evaluate the association of various ecological microbial communities, such as soil, water and rhizosphere. Presented here is a protocol on how to use the WGCNA algorithm to analyze different co-occurrence networks that may occur in the microbial communities due to different ecological environments.
This method can effectively explore the interactive relationship or co-occurrence network of different microbial species in different environments. It provides detail on how to use the WGCNA algorithm to construct a co-occurrence network of microbiota. In addition, based on the results, this method assesses the differentiation of microbial relationships and composition between microbial consortia.
Here is the basic flow of the method. The composition and abundance data of microbiota are downloaded from the NCBI database or sequence data for your samples. First, open the RStudio software and install the WGCNA package.
Then, load the data and use the good samples genes function to check the correctness of the data. Check for outliers and store samples that meet our requirements. When the check result is true, continue to the next step.
Save the result. Use the picks off threshold function to calculate the scale free index, R squared, of the data under different power values. Visualize the results.
When the scale free index is closer to one, the network structure is closer to the scale free network. When the scale free index, R squared, is greater than 0.9, select the power value. Finally, use the same method to analyze the rest of the microbiome data.
First, use the adjacency function to construct a symbolic co-occurrence network. In addition, use the TOM similarity function to develop a topological, overlapping network. Second, use the Hclust function to perform hierarchical clustering and draw the resulting cluster tree.
Third, use the cutree dynamic function to perform dynamic branch cutting and use the min cluster size parameter to set the smallest module size. The smallest module size is usually set over 30. Fourth, calculate the microbial characteristic of each module.
Hierarchical clustering is performed according to the correlation coefficient, and modules with a height of less than 0.25 were merged to acquire the distribution of each module. Fifth, use the plot dendro and colors function to visualize the results. The assignment display diagram of the co-occurrence network module is obtained.
Then, rename the module colors and construct digital labels corresponding to the colors. Save them for use in subsequent parts. Finally, repeat the above process for other sets of data.
In this part, compare and analyze the two sets of data and conduct the preservation test. First, load the parameters and results of the two datasets saved in the previous steps. Then, set the module results of one data set as the reference group, another as the test group and perform the module comparison.
Next, calculate the values of the conservativeness, statistical parameters, Z summary and median rank to quantify the conservativeness between modules. Finally, visualize the results. Determine the network module that satisfies both the Z summary value, less than two, and the median rank value at the top.
This module is the most non-preserved module in the two microbiota data. Perform correlation analysis of the module membership. Set the module assignment results of two networks as the reference and the test group respectively.
The settings need to be the same as the preservation test. First, the KME value of each OTU was calculated in several candidate modules based on the results of the preservation test. Take the yellow module as an example.
Calculate the correlation coefficient of the KME value in the two yellow modules, then draw the correlation analysis diagram of the KME value in the module. Finally, according to the correlation coefficient in the figure, judge the conservativeness of the module in the two data set. Select the module with the smallest correlation coefficient.
First, use the export network to Cytoscape function to export the network to an edge and node list file that Cytoscape can read. Then, import the file into Cytoscape. Set the threshold to 0.5 and adjust other parameters as needed.
Finally, obtain a co-occurrence network of different microorganisms. In this article, the WGCNA algorithm is used to analyze the differences in three niches of the rice root system. Select the power value that satisfied the three networks that were close to the scale free network.
In the endosphere, rhizoplane, and rhizosphere microbial occurrence network, identify 23, 22 and 21 models respectively. Use the preservation test and correlation analysis to find the extremely non-conservative modules between each two niches of the rice root system. Construct a co-occurrence network for these three modules using different colors to represent different microbial phylum.
Proteobacteria, Actinobacteria, Bacteroidetes, Firmicutes and Verrucomicrobia dominated the three different microbial networks. Moreover, 17 core genera mainly regulate these networks. After watching this video, you should have a good understanding of how to perform a series of steps to use the WGCNA algorithm to analyze different co-occurrence networks that may occur in the microbial communities due to different ecological environments.
View the full transcript and gain access to thousands of scientific videos
Overview
This article details a protocol for analyzing microbial co-occurrence networks in different ecological niches of the rice root system using the Weighted Gene Co-expression Network Analysis (WGCNA) algorithm. The method enables researchers to explore interactions and co-abundance patterns among microbial species, identify core genera, and assess network differences across environments such as the endosphere, rhizoplane, and rhizosphere soil.
Key Study Components
Area of Science
- Microbial ecology
- Bioinformatics
- Plant-microbe interactions
Background
- The root microbiome is crucial for plant growth and environmental adaptation.
- Network analysis helps reveal interactions and co-occurrence patterns among microbial species.
- WGCNA is an R package widely used for weighted correlation network analysis.
- Understanding microbial networks can inform ecological response theories and environmental adaptation mechanisms.
Purpose of Study
- To provide a detailed protocol for constructing and analyzing microbial co-occurrence networks using WGCNA.
- To compare microbial community structures across different rice root niches.
- To identify core genera and non-conserved modules within microbial networks.
Methods Used
- Download microbial composition and abundance data from the NCBI database.
- Install and use the WGCNA package in RStudio for network construction.
- Check data quality, identify outliers, and select appropriate samples.
- Determine optimal power values for scale-free network topology.
- Construct adjacency and topological overlap matrices (TOM).
- Perform hierarchical clustering and dynamic branch cutting to define modules.
- Visualize module assignments and merge similar modules.
- Conduct preservation tests and correlation analyses between modules from different datasets.
- Export network data to Cytoscape for visualization.
Main Results
- Constructed co-occurrence networks for the endosphere, rhizoplane, and rhizosphere soil of rice roots.
- Identified 23, 22, and 21 modules in the endosphere, rhizoplane, and rhizosphere, respectively.
- Detected non-conserved modules between different niches using preservation tests and correlation analysis.
- Proteobacteria, Actinobacteria, Bacteroidetes, Firmicutes, and Verrucomicrobia were dominant in the networks.
- Seventeen core genera were found to play key regulatory roles in the networks.
Conclusions
- The WGCNA-based protocol effectively reveals differences in microbial co-occurrence networks across rice root niches.
- Core genera and non-conserved modules can be identified, providing insights into microbial community adaptation.
- This approach supports the study of microbial ecological responses to environmental disturbances.
Frequently Asked Questions
Executive Industry Relevance
Weighted correlation network analysis (WGCNA) enables biopharma teams to dissect complex microbial community interactions, supporting predictive modeling of environmental or host-associated microbiomes. This approach enhances mechanistic de-risking and target validation by quantifying network divergence and identifying core regulatory taxa across distinct ecological niches. Integrating such network analytics into discovery pipelines informs translational strategies and portfolio prioritization for microbiome-targeted interventions.
Strategic Applications in Biopharma R&D
Early Discovery & Target Validation
- Enables interrogation of microbial interaction networks to clarify ecological drivers and regulatory taxa.
- Supports biological de-risking by quantifying divergence and conservation of microbial modules across environments.
- Facilitates predictive confidence in target selection by identifying core genera with regulatory influence.
Screening & Assay Development
- Prepares validated microbial community datasets for downstream functional screening or perturbation studies.
- Standardizes network module identification, supporting reproducibility and quantitative comparison across conditions.
- Enables scalable analysis of co-abundance patterns for robust assay development.
Translational & Preclinical Research
- Aligns microbial network features with disease-relevant or environmental adaptation models when supported by data.
- Supports continuity from ecological discovery to preclinical validation of microbiome-based interventions.
- Provides mechanistic insights for risk-adjusted advancement of microbiome targets.
Pipeline & Workflow Integration
WGCNA-based network analysis fits within the discovery-to-preclinical continuum, enabling hypothesis testing, target prioritization, and translational alignment for microbiome R&D.
- Discovery Biology: Quantifies network divergence and module conservation to clarify microbial community structure and function.
- Screening: Delivers reproducible module assignments and quantitative outputs for comparative analysis.
- Analytics: Provides statistical measures (e.g., Z summary, median rank) to benchmark network preservation and module distinctiveness.
- Translational Research: Supports mapping of core genera to functional or disease-relevant outcomes when data permit.
- Enterprise Reuse: Establishes a reusable analytic framework for diverse microbiome datasets and environments.
Operational & Enterprise Impact
- Scientific Value: Increases predictive confidence and reduces mechanistic ambiguity in microbiome target validation.
- Operational Value: Standardizes network analysis workflows for reproducibility and scalability across projects.
- Strategic Value: Informs go/no-go decisions and capital allocation by quantifying network divergence and core taxa.
- Portfolio Impact: Enables risk-adjusted prioritization of microbiome targets and interventions.
Implementation Considerations
- Requires expertise in R, network analysis, and microbial ecology.
- Needs computational infrastructure for large-scale data processing and visualization (e.g., Cytoscape).
- Demands cross-team standardization of data quality checks and module assignment parameters.
- Adaptation across different host or environmental microbiomes may require parameter tuning.
- Interpretation of network modules should be grounded in ecological or functional context supported by data.
Biopharma FAQs
Chapters in this video
0:06
Introduction
0:46
I. Data Download
1:05
II. Optimal Power Value Determination
2:38
III. Construction of a Co-Expression Network and Module Identification
4:25
IV. Preservation Test
5:44
V. Correlation Analysis of the Module Membership
6:55
VI. Analysis of the Microbial Flora in the Microbial Differential Network Module
7:43
Resuts: Analysis of Differences in Rice delavayi Root Microbial Communities Based on Weighted Co‐Expression Network
9:18
Conclusion
