μODE — FMT engraftment & C. difficile colonization-resistance: reproducibility archive

Abstract

This archive contains all analysis code, derived data, run outputs, and the figure pipeline needed to reproduce the figures, tables, and quantitative results of the manuscript: Mechanistic prediction of fecal microbiota transplant engraftment and Clostridioides difficile colonization resistance with μODEFabio Cumbo, Jayadev Joshi, Bryan Raubenolt, Natalie Katzenmeyer, Jan Claesen, Daniel Blankenberg μODE is an automated, open-source framework that converts metagenome-assembled genomes (MAGs) into dynamic flux-balance (dFBA) simulations of community metabolism, coupled to a composable ecology layer (pH/short-chain-fatty-acid inhibition, bile-acid transformation, spore dormancy, and time-varying antibiotic pharmacokinetics). The framework itself is released separately at https://github.com/cumbof/muODE (MIT license). This archive holds the paper-specific analyses only; the μODE package and its genome-scale models are referenced rather than duplicated here (see Data not included below). The figure scripts are standalone R (ggplot2) and read only from this archive. prepare_data.R documents how each tidy figures/data/*.csv was aggregated from the raw run outputs in results/. Requirements. R (≥ 4.2) with ggplot2, dplyr, readr, forcats, scales, grid, and patchwork; cairo graphics devices for PDF/PNG export. The simulation and reconstruction code requires the μODE package (see above) and its dependencies (Python ≥ 3.10, COBRApy, an LP solver), gapseq for reconstruction, and MICOM for the community-objective test. Data not included (and where to get it) Large genome-scale models and raw sequencing data are not duplicated here: μODE package + the 139 reconstructed genome-scale metabolic models and run configs: https://github.com/cumbof/muODE Clinical FMT metagenomes (longitudinal rCDI "Subject S"): NCBI SRA BioProject PRJNA705895. Clark et al. 2021 defined-community benchmark: Nat. Commun. 12:3254 (data with the original paper). Buyanbadrakh et al. 2026 36-strain spent-medium interactions: Zenodo 17456178, PRIDE PXD070029, MassIVE MSV000102556. In vivo mouse cecal bile-acid measurements used for quantitative validation: Theriot et al. (Nat. Commun. 2014; mSphere 2016). Determinism & provenance The μODE dynamic engine is deterministic: identical inputs, parameters, and time steps reproduce a run exactly. Run logs and the exact configurations are under results/. All scripts are provided as used; absolute paths inside them document the original compute environment and should be adapted to a local checkout. License & citation Code and derived data in this archive are released under the MIT License, consistent with the μODE software. If you use this archive, please cite the manuscript above and this Zenodo record.