Microbiomes harbor intricate associations, or co-occurrences, between member taxa that may be critical to the assembly and function of the microbial community. Identifying and characterizing these associations from metagenomic data may offer insight into disease progression and prevention. The organization of the microbiome is thought to be a singular assortative network, where interactions between taxa can readily be clustered into segregated and distinct communities. However, identifying distinct microbial communities is a statistical challenge due to the complexity of microbiome data. Microbiome Community Detector (MiCoDe) is a web tool that can be used to investigate the community structure of co-occurrence networks estimated from microbiome data. MiCoDe fits either a Bayesian nonparametric weighted stochastic block model (WSBM) or a Bayesian nonparametric weighted stochastic infinite block model (WSIBM) to a fully connected network represented by a graph with weighted edges.
1. Upload CSV
2. Run Analysis
3. Download Results
Users upload a taxonomic abundance table (in CSV format) with rows and columns corresponding to samples and taxa, respectively. Before running the analysis, users have several arguments to choose from including data preprocessing (MCLR, CLR, Compositional), taxon-taxon association measures (SPR, Spearman, or Pearson correlation coefficients), and the number of communities (select ‘automatic’ to use WSIBM; or prespecify a value from 2 to 20 to use WSBM). The output is generated in the form of an interactive network separated by communities allowing users to dynamically filter edges by correlation magnitude. By placing your cursor over a node or edge, additional information will be displayed such as the correlation coefficient between two taxa or the number of edges connected to a particular taxon. For better visualization of a particular community, you can filter out other communities from the network. Simply click on the community names in the legend to remove those communities from the network. Click a second time to add them back to the network. Lastly, the output is saved for later use. Use the link in the output to access your results later without having to re-run the analysis.
Lutz, Kevin C., Shengjie Yang, Tejasv Bedi, Michael L. Neugent, Nikita Madhavaram, Bo Yao, Xiaowei Zhan, Nicole J. De Nisco, and Qiwei Li. "MiCoDe: A web tool for performing microbiome community detection using a Bayesian weighted stochastic block model." Bioinformatics (2025): btaf384.
Link: https://academic.oup.com/bioinformatics/article/41/7/btaf384/8180607