Nonlinear network inference reveals two independent axes of ecological organisation in the rumen microbiome

FANCY (Frequency And Nonlinear Correlation HYbrid)Supplementary data and reproducibility deposit 1. Main_Supplement_Fancy.pdf: the full manuscript and supplementary information (Figures S1–S22, Tables S1–S7). Read this first for context.2. README.md: A sectioned guide to every file in this deposit and the figure or table it supports.3. The rendered html reports, each embeds the R code and intermediate outputs showing how the corresponding analysis was produced. Together they document the reproducibility of the module-validation, cross-method comparison, and network-robustness analyses. They do not re-run every upstream step: MAG assembly and functional annotation, the core FANCY network inference (provided as an R package at https://github.com/wala-github/Fancy), and the interactive Cytoscape GLay clustering are external to these reports. Citation Lai W., Leu A., Roehe R., Pope P. B., Hvidsten T. R. Nonlinear network inference reveals two independent axes of ecological organisation in the rumen microbiome. (Manuscript, 2026). Overview FANCY integrates k-nearest-neighbour mutual information, MRNET feature selection, and distance-correlation stability into a hybrid network-inference framework for microbiome association analysis. Applied to 2,178 metagenome-assembled genomes (MAGs) from 321 beef steers, FANCY identified three dominant modules associated with: Rumen Community Type (RCT) structure Host-genetic / breed-associated structure Methanogenesis pathway concentration This deposit provides the network files, validation datasets, intermediate outputs, and summary tables supporting the analyses reported in the manuscript. Contents The deposit includes: Hybrid edge tables and derived network files Network robustness-analysis summary tables Module-validation datasets and enrichment results Comparator-network outputs (SPIEC-EASI, SparCC, Pearson) Intermediate files required to verify FANCY network construction Functional annotation resources used in downstream analyses Detailed descriptions of individual files are provided in the accompanying file index. Software environment Analyses were performed in R using the FANCY package together with standard network-analysis and statistical libraries, including: igraph data.table SpiecEasi minet knnmi energy Related resources FANCY: https://doi.org/10.5281/zenodo.20003547 GitHub repository:https://github.com/wala-github/Fancy License Data files are released under CC BY 4.0. Software components, where included, are released under the MIT License. Contact For questions regarding this deposit or the associated manuscript, please contact the corresponding author listed in the publication. Acknowledgments The RCT assignments and methanogenesis reference gene sets used in downstream validation analyses originate from previously published studies and remain subject to the citation requirements of their original publications.