This repository provides resources for performing microbiome multi-omics analyses in the Knight Lab.
The resources provided here focus on mmvec and joint-rpca.
Multi-omics integration involves combining data from different biological layers (i.e., 16S, metagenomics, metatranscriptomics, metabolomics, etc).
| Feature | mmvec | joint-rpca |
|---|---|---|
| Objective | Estimating microbe-metabolite interactions through their co-occurence probabilities using neural networks. | Identifying features across the multiple 0omics types which separate jointly across the samples using a dimensionality reduction method. |
| Input Format | Microbial feature table, metabolite feature table, and metadata. | Paired feature tables in compositional format. |
| Number of Omics Layers | Limited to two -omics layers. | Can have any number of -omics layers. |
| Ouputs | Covariance matrix between -omics layer 1 and -omics layer 2, conditional biplot which can be input into Emperor, evaluations of probability model performance. | Distance matrix represents the pairwise distances between samples in the reduced compositional feature space derived from the joint factorization of multiple -omics datasets, biplot visualization, evaluations of dimensionality reduction model performance |
| Ordination/QIIME2 Biplot | Points are -omic layer 1 (i.e., microbes) and arrows are -omic layer 2 (i.e., metabolites). | Points are all samples and arrows can be features from any -omic layer (i.e., 16S, metagenomic, metabolomic, etc.). |
To use mmvec and joint-rpca, functional QIIME2 environments have been created and optimized for Barnacle2. To install these environments:
# Clone the repository
git clone https://github.com/yangchen2/multiomics.git
cd multiomics
# Install mmvec environment
mamba env create -f mmvec_qiime2-2020.6_barnacle2.yml
mmvec paired-omics --help
# Install joint-rpca environment
mamba env create -f joint-rpca_qiime2-2022.11_barnacle2.yml
qiime gemelli joint-rpca --help
- Additional tutorials and walkthroughs for setting up mmvec and joint-rpca analyses.
- Example scripts and Jupyter notebooks for analyzing multi-omics datasets.
- Strategies for interpreting outputs and integrating them into broader studies.
- Visualization scripts for mmvec and joint-rpca outputs.
- This repository has been adapted from https://github.com/ahdilmore/multiomics.
- For Hazel Dilmore's previous code review presentation, refer to this link.
If you are a Knight Lab member and have additional resources to contribute, please email [email protected].