This repository contains a Snakemake pipeline to conduct metabolomics data analysis comparing two groups of samples in a robust manner, handling the random component on every step of the analysis.
Most of the code to handle the metabolomics datasets is based on the metabotk python library (available at https://github.com/matteobolner/metabotk).
Variants of this pipeline have been used to analyze metabolomics data in the following publications:
- High-throughput untargeted metabolomics reveals metabolites and metabolic pathways that differentiate two divergent pig breeds (Published)
- Description of metabolic differences between castrated males and intact gilts obtained from high-throughput metabolomics of porcine plasma (Accepted for publication)
Edit the config.yaml file to define the dataset and analysis parameters. The metabolomics dataset must be in excel format and contain the following sheets:
- Sample metadata
- Chemical annotation (metabolite metadata)
- Metabolite abundance data
The summarized structure of this pipeline can be seen below (rulegraph.svg). For the full pipeline including the repeated steps to account for multiple random seeds, see dag.svg
- Missing data imputation
- Remove outlier values and samples/metabolites with too many missing values (> 25% by default)
- Impute missing values with MICE predictive mean matching (pmm)
- Normalization (optional)
- Removal of confounding effects (OLS regression)
- Feature selection
- Boruta (Random Forest)
- Differential metabolite analysis
- Univariate analyses (Mann-Whitney, ROC AUC)
- Multivariate analyses (Correlation networks) - not added to this repo yet
If you use this workflow and want to cite it, you can cite our paper: High-throughput untargeted metabolomics reveals metabolites and metabolic pathways that differentiate two divergent pig breeds