Giotto contains several functions that contain wrappers to Python code and thus requires an environment containing Python. Utilizing the functionality of the reticulate package, Giotto contains a function which sets up a miniconda environment and installs the required Python packages within that environment. Once this function, installGiottoEnvironment, has been run, Giotto will automatically default to utilizing this environment.
# Ensure Giotto Suite is installed
if(!"Giotto" %in% installed.packages()) {
pak::pkg_install("drieslab/Giotto")
}
library(Giotto)
# Ensure Giotto Data is installed
if(!"GiottoData" %in% installed.packages()) {
pak::pkg_install("drieslab/GiottoData")
}
library(GiottoData)
# Ensure the Python environment for Giotto has been installed
genv_exists <- checkGiottoEnvironment()
if(!genv_exists){
# The following command need only be run once to install the Giotto environment
installGiottoEnvironment()
}The function installGiottoEnvironment two particular arguments that are most useful for reinstallation, if necessary:
Note that, by default, installGiottoEnvironment installs a specific version of Python and each required package. At the time of this tutorial"s creation, the following versions are utilized:
If different versions of Python or packages are necessary for a workflow, Giotto may be installed accordingly. Ensure that all required packages, which have been listed above, are accounted for when installing. Simply specify the desired version numbers for each package within a vector, and provide that vector to the packages_to_install argument within installGiottoEnvironment.
Note that machine type is not relevant when providing packages_to_install to installGiottoEnvironment; this function will identify the OS in use and install/not install packages (i.e. python.app) accordingly.
### Note that the following code has been provided to indicate how to install
### Giotto with customized Python and Python package versions. It has been
### intentionally commented out so that it will not run and overwrite the
### default versions unless deliberately edited.
### new_pkg_versions <- c("pandas == 1.4.4",
### "networkx == 2.6.3",
### "python-igraph == 0.9.6",
### "leidenalg == 0.8.7",
### "python-louvain == 0.15",
### "scikit-learn == 0.24.2",
### "python.app == 2")
###
### ############################
### # If altering the original Giotto Installation is not desired, DO NOT
### # run the following command as written.
### ############################
### installGiottoEnvironment(packages_to_install = new_pkg_versions,
### python_version = "3.8") # Default is 3.10.2If using reticulate"s default miniconda path to create an environment is undesirable, the Giotto environment may be created within an existing anaconda/miniconda environment by specifying the mini_install_path argument:
installGiottoEnvironment(mini_install_path = "path/to/conda")If not provided, it is chosen by reticulate::install_miniconda(). Please note the required input format: - Correct format: mini_install_path = “C:/my/conda/lives/here” OR “C:\my\conda\lives\here” - INCORRECT formats: mini_install_path = “C:/my/conda/lives/here/” AND “C:\my\conda\lives\here\”
Unexpected behavior could arise if force_miniconda is set to TRUE when mini_install_path is specified and encompasses a non-reticulate environment, as this prompts a reticulate miniconda installation.
Note that the installation of all aforementioned packages is necessary for the full functionality of Giotto. A .yml file is provided in the repository for convenience of alternative installation methods. If the desired environment is not named “giotto_env”, Giotto will be unable to automatically detect the conda environment, so it must be specified within a workflow. To use a specific, non-default named Conda environment, the path to a system-specific python executable within that environment must be provided to createGiottoInstructions. This will direct reticulate to activate and utilize that environment within that R session. See How to Create a Giotto Object for more details.
Giotto makes use of the following Python packages (and their respective dependencies) for full functionality:
Here is a brief troubleshooting workflow to investigate if reticulate can access them.
Note that “community” and “sklearn” are aliases of “python-louvain” and “scikit-learn”, respectively.
# Creating Giotto Instructions without specifying a Python path will make
# reticulate activate the default Giotto environment.
instructions <- createGiottoInstructions()
# Extract python path information
python_path <- instructions$python_path
# Make reticulate iteratively check for the packages
pkg_check <- function(){
py_pkgs = c("pandas", "networkx", "igraph", "leidenalg", "community", "sklearn",
"python.app")
py_pkg_error = character()
test_availability = TRUE
for (i in py_pkgs){
if(i == "python.app" & Sys.info()[["sysname"]] != "Darwin"){
# If the machine OS is not OSX (Mac), break out of the loop
# Otherwise, also check for python.app
break
}
test_availability <- reticulate::py_module_available(i)
if(!test_availability) {py_pkg_error <- c(py_pkg_error,i)}
}
if(test_availability){
cat("All Python packages for Giotto are accessible at environment:\n", python_path)
}else{
for (x in py_pkg_error)
cat(x, "was not found within environment:\n", python_path, "\n")
}
return(py_pkg_error)
}
pkg_check()
# Restart the R session, while maintaining workspace variables.
# If using RStudio, the following command will do exactly that:
.rs.restartR()
# Direct reticulate to use Python within the Giotto Environment
reticulate::use_python(python_path)
# Check if packages exist again. Ensure function from above code block is defined.
missing_packages <- pkg_check()
retry_install <- length(missing_packages) > 0
if(retry_install){
# Attempt to reinstall all packages.
pkgs_w_versions <- c("pandas == 1.5.1",
"networkx == 2.8.8",
"python-igraph == 0.10.2",
"leidenalg == 0.9.0",
"python-louvain == 0.16",
"python.app == 1.4",
"scikit-learn == 1.1.3")
python_version <- "3.10.2"
py_pkgs <- c("pandas", "networkx", "igraph", "leidenalg",
"python-louvain", "scikit-learn", "python.app")
if(Sys.info()[["sysname"]] != "Darwin"){
pkgs_w_versions <- pkgs_w_versions[!grepl(pattern = "python.app", x = pkgs_w_versions)]
py_pkgs <- py_pkgs[!grepl(pattern = "python.app", x = py_pkgs)]
}
env_location <- reticulate::py_discover_config()$pythonhome
partial_path_to_conda <- paste0(reticulate::miniconda_path(),"/envs/giotto_env")
py_lou <- pkgs_w_versions[grepl(pattern = "python-louvain", x = pkgs_w_versions)]
pip_packages <- c("smfishhmrf", py_lou)
pkgs_w_versions <- pkgs_w_versions[!grepl(pattern = "python-louvain", x = pkgs_w_versions)]
if(.Platform[["OS.type"]] == "unix") {
conda_full_path = paste0(partial_path_to_conda, "/bin/conda")
# Remove all previous installations
reticulate::conda_remove(envname = env_location,
packages = py_pkgs,
conda = conda_full_path)
# Reinstall
reticulate::conda_install(packages = pkgs_w_versions,
envname = env_location,
method = "conda",
conda = conda_full_path,
python_version = python_version)
# Reinstall with pip
reticulate::conda_install(packages = pip_packages,
envname = env_location,
method = "conda",
conda = conda_full_path,
pip = TRUE,
python_version = python_version)
}
else if(.Platform[["OS.type"]] == "windows"){
conda_full_path = paste0(partial_path_to_conda,"/","condabin/conda.bat")
# Remove all previous installations
reticulate::conda_remove(envname = env_location,
packages = py_pkgs,
conda = conda_full_path)
# Reinstall
reticulate::conda_install(packages = pkgs_w_versions,
envname = env_location,
method = "conda",
conda = conda_full_path,
python_version = python_version,
channel = c("conda-forge", "vtraag"))
# Reinstall with pip
reticulate::conda_install(packages = pip_packages,
envname = env_location,
method = "conda",
conda = conda_full_path,
pip = TRUE,
python_version = python_version)
}
}If this does not fix the issue at hand, here are some potential action items:
R version 4.3.2 (2023-10-31)
Platform: x86_64-apple-darwin20 (64-bit)
Running under: macOS Sonoma 14.3.1
Matrix products: default
BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
LAPACK: /Library/Frameworks/R.framework/Versions/4.3-x86_64/Resources/lib/libRlapack.dylib; LAPACK version 3.11.0
locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
time zone: America/New_York
tzcode source: internal
attached base packages:
[1] stats graphics grDevices utils datasets methods base
other attached packages:
[1] Giotto_4.0.2 GiottoClass_0.1.3
loaded via a namespace (and not attached):
[1] SummarizedExperiment_1.32.0 gtable_0.3.4 rjson_0.2.21
[4] xfun_0.42 ggplot2_3.4.4 Biobase_2.62.0
[7] lattice_0.22-5 vctrs_0.6.5 tools_4.3.2
[10] bitops_1.0-7 generics_0.1.3 parallel_4.3.2
[13] stats4_4.3.2 tibble_3.2.1 fansi_1.0.6
[16] colorRamp2_0.1.0 pkgconfig_2.0.3 Matrix_1.6-5
[19] checkmate_2.3.1 data.table_1.15.0 S4Vectors_0.40.2
[22] lifecycle_1.0.4 GenomeInfoDbData_1.2.11 compiler_4.3.2
[25] GiottoUtils_0.1.5 munsell_0.5.0 terra_1.7-71
[28] codetools_0.2-19 GenomeInfoDb_1.38.6 htmltools_0.5.7
[31] RCurl_1.98-1.14 yaml_2.3.8 pillar_1.9.0
[34] crayon_1.5.2 SingleCellExperiment_1.24.0 DelayedArray_0.28.0
[37] magick_2.8.2 abind_1.4-5 gtools_3.9.5
[40] tidyselect_1.2.0 digest_0.6.34 dplyr_1.1.4
[43] rprojroot_2.0.4 fastmap_1.1.1 grid_4.3.2
[46] here_1.0.1 colorspace_2.1-0 cli_3.6.2
[49] SparseArray_1.2.3 magrittr_2.0.3 S4Arrays_1.2.0
[52] utf8_1.2.4 withr_3.0.0 backports_1.4.1
[55] scales_1.3.0 rmarkdown_2.25 XVector_0.42.0
[58] matrixStats_1.2.0 reticulate_1.35.0 GiottoVisuals_0.1.4
[61] png_0.1-8 SpatialExperiment_1.12.0 evaluate_0.23
[64] knitr_1.45 GenomicRanges_1.54.1 IRanges_2.36.0
[67] rlang_1.1.3 Rcpp_1.0.12 glue_1.7.0
[70] BiocGenerics_0.48.1 rstudioapi_0.15.0 jsonlite_1.8.8
[73] R6_2.5.1 MatrixGenerics_1.14.0 zlibbioc_1.48.0