Tutorials (original) (raw)
Precorded talks
- MOFA overview: precorded talk for the VIB workshop (Belgium, 2021), includes the model overview, intuition and a brief discussion of the CLL application.
- Overview of single-cell multi-omics data integration: precorded talk for a webinar, includes brief discussion on CITE-seq and 10x Multiome applications.
Getting started using R
- Training a MOFA model in R: using simple simulated data
- Downstream analysis in R: using simple simulated data
- Gene set enrichment analysis: demonstrates how to do gene set enrichment analysis in R.
- Demonstration of the stochastic inference algorithm: this is only useful for very large data sets and when having access to GPUs.
Case examples using real data (in R)
- (authors’ favourite) Analysis of chronic lymphocytic leukaemia cohort for personalised medicine: a bulk multi-omics data set. Figure 2 and 3 of the [MOFA paper (https://www.embopress.org/doi/full/10.15252/msb.20178124#msb178124-fig-0002).
- Integrative analysis of the Chromium Single Cell Multiome ATAC + Gene Expression assay: this is the result of a collaboration between the MOFA team and the 10x Genomics R&D team to provide a downstream analysis pipeline for the 10x Multiome kit.
- Analysis of multi-modal microbiome data: we demonstrate how to systematically integrate viral, fungal and bacterial sequence data. Manuscript published in mSystems
- Analysis of a time course scRNA-seq data set using the multi-group framework: Figure 2 of the MOFA+ paper. Demonstrates the multi-group functionality and how to train a MOFA model from a Seurat object.
- Integration of scNMT-seq data (single-cell multi-omics): Figure 4 of the MOFA+ paper. Demonstrates the simultaneous multi-view and multi-group functionality using the multi-modal mouse gastrulation atlas.
- Integration of SNARE-seq data (single-cell multi-omics). Demonstrates how MOFA can be used for the analysis of paired scRNA+scATAC data (from the same cell) from a Seurat object. This data set is very noisy and the results are not fantastic, we suggest you have a look at the 10x Multiome vignette instead.
All .Rmd files for the tutorials can be found here
Getting started using Python
MOFA is interfaced through muon:
- Training MOFA with muon, CLL: the multimodal CLL dataset in the MuData format.
- Integration of multimodal single-cell data: shows how muon can be used to train and explore the model together with the mofax package.
- Training a model with mofapy2: a Jupyter notebook demonstrating how to train a MOFA model using simple simulated data. It uses stand-alone mofapy2 to train the model and mofax for downstream analysis.
- Training a model on AnnData: demonstrates how to train a MOFA model on scRNA-seq data stored in AnnData format.
- Downstream analysis in Python: demonstrates how to perform the downstream analysis of a MOFA model trained on scRNA-seq data, using mofax.
Template scripts
We provide template scripts to train your model: