Prepare the python environment — PrepareEnv • scop

Prepare the python environment by installing the required dependencies and setting up the environment.

Usage

PrepareEnv(
  envname = NULL,
  conda = "auto",
  miniconda_repo = "https://repo.anaconda.com/miniconda",
  version = if (is_windows()) "3.11-1" else "3.10-1",
  force = FALSE,
  modules = NULL,
  pip_options = character(),
  components = "python",
  cores = NULL,
  verbose = TRUE,
  configure_runtime = TRUE,
  ...
)

Arguments

envname: Conda-compatible Python environment. If NULL, use the current environment selection. PrepareEnv() selects a separate module profile when required.
conda: The path or command name of a conda-compatible executable (conda, mamba, or micromamba). Use "auto" to allow automatically finding an appropriate environment manager. If "micromamba" is requested and micromamba is not available on PATH, a package-managed micromamba is downloaded automatically.
miniconda_repo: Repository URL for miniconda. Default is https://repo.anaconda.com/miniconda.
version: The Python version. Default is "3.10-1" on macOS and Unix and "3.11-1" on Windows. Existing compatible environments are reused; trajectory package pins are selected for the actual Python 3.10/3.11/3.12 profile.
force: Whether to force recreation of the environment. If TRUE, the existing environment will be removed and recreated.
modules: Optional Python dependency modules to install in addition to the default scientific stack. If NULL or omitted in PrepareEnv(), the default environment is installed. The default excludes "phate", "magic", "cell2location", "cell2fate", "commot", "sccoda", and "scomm" because these workflows use long-running or incompatible dependency stacks. Cell2fate and COMMOT are prepared in standalone "cell2fate_env" and "commot_env" environments with Python 3.9. SCENIC+ uses standalone "scenicplus_env" or the smaller "scenicplus_io_env" for .h5mu imports. "scenic" is also excluded from the default environment and is prepared in "scenic_env" by default because SCENIC requires an older Python/numpy stack. PHATE and MAGIC are prepared explicitly in "scprep_env" with pandas 2.0.3, separately from the default scientific stack. scCODA and scOMM use "tensorflow_env" to keep TensorFlow dependencies separate from JAX. "scmalignantfinder" is prepared in a standalone "scmalignantfinder_env" because its upstream package pins an older scanpy/anndata stack. On Windows, the default also excludes "scvi", "glue", and "multimap" because those upstream stacks are more reliable when requested explicitly for method-specific workflows. CellOracle uses an explicit standalone "celloracle_env" profile with Python 3.10 and official CellOracle 0.20.0; Windows is unsupported.
pip_options: Additional command line arguments to be passed to uv/pip when installing pip packages.
components: Components to prepare. Supported values are "python", "r", and "all"; "all" expands to both Python and R. The default, "python", preserves the existing Python-only behavior. The R component collects packages declared through check_r() and checks/installs them, including optional workflow dependencies not listed in DESCRIPTION. CHOIR is R-only, so PrepareEnv(modules = "choir") prepares its optional R implementation without creating a Python environment.
cores: Number of workers for R-package installation. Use NULL (the default) to let pak select its worker count automatically.
verbose: Whether to print the message. Default is TRUE.
configure_runtime: Whether to select the prepared interpreter for embedded Python and configure the R process. Set to FALSE for subprocess-only use.
...: Arguments passed to other functions.