Installation

The core of MOFA is implemented in the Python package mofapy2, but we recommend to use the R package MOFA2 which provides an interface to train a MOFA model with R and run the downstream analysis and takes care of setting up all python dependencies. Alternatively, if you prefer to use Python the package mofax can be used for downstream analysis in Python, see also our FAQ section.

Stable release (easiest)

You can install the stable release from Bioconductor (link):

if (!requireNamespace("BiocManager", quietly = TRUE))
    install.packages("BiocManager")

BiocManager::install("MOFA2")

Developmental version

To use the latest features of MOFA you can install the software from GitHub:

devtools::install_github("bioFAM/MOFA2", build_opts = c("--no-resave-data --no-build-vignettes"))

You will also need to make mofapy2 available to R — see below. If you’d like the development version of mofapy2 to match, install it directly from GitHub:

reticulate::py_require("git+https://github.com/bioFAM/mofapy2")

Notes on the connection of R to Python

The connection between R and Python is done via reticulate. There are three ways to make mofapy2 available to the R package:

Let reticulate provision it (simplest, requires reticulate >= 1.41). Declare the dependency before training, and reticulate sets up an isolated Python environment — downloading a suitable Python interpreter if none is available:
```
reticulate::py_require("mofapy2")
MOFAobject <- run_mofa(MOFAobject)
```
Let basilisk handle it. run_mofa(MOFAobject, use_basilisk = TRUE) uses a dedicated, version-pinned Python environment, created the first time you call it this way. It is also the option to choose if you use MOFA2 alongside other R packages with conflicting Python dependencies, since basilisk runs Python in a separate process.
Use an existing Python installation. Install the mofapy2 package and its dependencies manually with pip install mofapy2 (from the Unix terminal), then select that installation with reticulate::use_python() (or reticulate::use_condaenv()) - this option needs the most configuration.

Note that the connection of R and python is the source of most problems when running MOFA, see our troubleshooting page or reach us if you have issues.

Using MOFA2 with older R versions

We recommend using R (>= 4.0) with MOFA2. If you want to use it with older R versions, you can install MOFA2 as

remotes::install_github("bioFAM/MOFA2", ref = "R36", build_opts = c("--no-resave-data --no-build-vignettes"))

Note, that this is only maintained intermittently and you will need to manually install the python package as described above and possibly configure the reticulate interface.

Installation using Docker image

If you use Docker, you can build an image using the provided Dockerfile:

docker build -t mofa2 .

You will then be able to use R or Python from the container.

docker run -ti --rm -v $DATA_DIRECTORY:/data mofa2 R
#                   ^
#                   |
#                    use `-v` to map a folder on your machine to a container directory

The command above will launch R with MOFA2 and its dependencies installed while mounting $DATA_DIRECTORY to the container.

You can also pull the pre-build image from dockerhub.