This repository stores the code and data for comparing microbial co-occurrence networks across treatment groups (AGP, PFA, Control) using chicken cecum metagenomic data at days 14, 21, and 35.
pip install -r requirements.txt
python src/main.py
data/raw/ - abundance tables, taxonomy, metadata
data/processed/ - GraphML network files
src/ - analysis code
main.py - runs the full pipeline
metrics.py - network topology metrics
keystone.py - keystone taxa identification
robustness.py - network robustness simulations
rf.py - random forest classification
plotting.py - figures
utils.py - helpers
figures/ - output plots
output/ - results and notebooks
- Computes network topology metrics (modularity, clustering, density, etc.)
- Identifies keystone taxa by centrality measures
- Tests network robustness via node removal simulations
- Classifies treatment groups using random forest on keystone taxa abundance
MIT