Setting up LaminR

This guide provides more detailed instructions for how to set up {laminr}.

Installing {laminr}

The {laminr} package can be installed from CRAN:

install.packages("laminr")

Additional packages

Some functionality in {laminr} requires additional packages. To install all suggested dependencies use:

install.packages("laminr", dependencies = TRUE)

If you choose not to install all additional packages you will be prompted to do so as needed.

Installing the development version

The current development version of {laminr} can be installed from GitHub using:

if (!requireNamespace("laminr", quietly = TRUE)) {
  install.packages("remotes")
}
remotes::install_github("laminlabs/laminr")

This version may have bug fixes and new features but could also be unstable so is not recommended for most users.

Installing Python lamindb

Using {laminr} requires the Python lamindb package to be available.

Using the included Python environment

The recommended way to use {laminr} is to connect to the included r-lamindb Python environment. This can be created using:

laminr::install_lamindb()

This should be run before the first time you load the {laminr} library. The {reticulate} package will then be told to use this environment when {laminr} is loaded.

Adding additional packages

If you want to add additional packages to the environment, these can be specified using the extra_packages argument. For example, to add the bionty package:

laminr::install_lamindb(extra_packages = "bionty")

Example of Python packages providing additional registries that may be used in your instance:

  • bionty - Basic biological entities, coupled to public ontologies

  • clinicore - Basic clinical entities

  • omop - OMOP Common Data Model

  • wetlab - Basic wetlab entities

Installing additional Python packages also be useful if you want to use another R package which expects a particular Python module to be installed.

Adding lamindb to another environment

To install lamindb into another environment, set the envname argument:

laminr::install_lamindb(envname = "your-env")

Be aware that there may be dependency conflicts with packages already installed in the environment and installing lamindb may cause package versions to change.

Setting the Python environment

Running library(laminr) will tell {reticulate} to use the included r-lamindb environment (if it has not already connected to another environment).

If you prefer to not load the whole package, and instead call functions using laminr::, you should tell {reticulate} which environment to use by running reticulate::use_virtualenv("r-lamindb").

Using another environment

In some cases you may prefer to use another Python environment. This may be the case if you use another R package which requires a Python module or if you are managing your own Python environment.

Specifying another Python environment to use should be done before loading {laminr} (or {reticulate}) using one of these methods:

  • Loading another R package that sets the Python environment

  • Using reticulate::use_virtualenv(), reticulate::use_condaenv() or reticulate::use_python()

  • Setting the RETICULATE_PYTHON or RETICULATE_PYTHON_ENV environment variables

Details of the current Python environment can be viewed using reticulate::py_config(). For more information about setting the active Python environment see vignette("versions", package = "reticulate") (also available here).

Logging in to LaminDB

If you have not used LaminDB before you will need to login to access public instances. To do this you will need a user API key which you can get by logging in to LaminHub and going to your user settings.

You can then login with:

laminr::lamin_login(api_key = "your_api_key")

Switching users

If you have already given an API key for a user you can log in as them by giving the user name:

laminr::lamin_login(user = "user_handle")

Setting a default instance

Using import_module("lamindb") will connect to the current default LaminDB instance. The default instance can be set using:

laminr::lamin_connect("<owner>/<name>")

Note that this must be called before attempting to connect to a default instance and cannot be changed once the default instance is connected without first calling lamin_disconnect().