Manual package installation¶

Attention

While it is fine to install the R-package in the way described here, we do not recommend this for general users and refers to the platform specific installation instructions instead.

Step by step guide¶

You can install the development version of cecelia like so:

Install github version¶

if (!require("remotes", quietly = TRUE))
  install.packages("remotes")
remotes::install_github("schienstockd/cecelia", Ncpus = 4, repos = "https://cloud.r-project.org")

After the install is finished, you must install the requirements for the shiny app.

Some packages must be installed from github repos, hence this manual step after the main package installation.
Install shiny app requirements¶
library(cecelia)
cciaAppRequirements(repos = "https://cloud.r-project.org")

Cecelia utilises a suite of flow cytometry packages, mainly from the RGLab, which needs to be installed from Bioconductor or github.

Install Bioconductor requirements¶
cciaBiocRequirements()

You must define a base directory where configuration files, models and the shiny app will be stored.

Define base directory¶
cciaSetup("/REPLACE/WITH/YOUR/PATH")
In case you get stuck at any point and have to restart R, you need to redefine the path you are working on
Restart Cecelia¶
# if you already installed flowCore during cciaBiocRequirements, otherwise comment out
library(flowCore)
Sys.setenv(KMP_DUPLICATE_LIB_OK = "TRUE")
library(cecelia)
cciaUse("/REPLACE/WITH/YOUR/PATH")

Cecelia depends on a conda environment which must be created.

There are multiple options available depending on how you would like to use the app:

image For image analysis on Desktop (default)

image-nogui For image processing without GUI

flow For flow cytometry analysis
Tip

Install miniconda for Apple Metal systems, see github issue.
Install miniconda¶
reticulate::install_miniconda()
Attention

For Apple Metal systems, you need to pre-create the conda environment otherwise it will use the wrong platform type. You might have to source conda first if the command cannot be found. reticulate::miniconda_path() will give you the conda path.
Pre-create conda environment¶
# if the conda command is not found
. /path/to/miniconda/etc/profile.d/conda.sh
CONDA_SUBDIR=osx-arm64 conda create -n r-cecelia-env python=3.9
Create conda environment¶
cciaCondaCreate(envType = "image")
Attention

If the previous command fails on PyQt5 on Apple Metal systems, preinstall napari.
Preinstall napari¶
cciaCondaCreate(preinstallNapari = TRUE)

Download models for deep-learning segmentation, tracking and others.

Download models¶
cciaModels()

Create shiny app in base directory.

Create shiny app.¶
cciaCreateApp()

Adjust config

You have to adjust the parameters in ~/path/to/cecelia/custom.yml to your system and download/install:

bioformats2raw

Adjust config.¶

default:
  dirs:
    bioformats2raw: "/path/to/bioformats2raw"
    projects: "/your/project/directory/"
  volumes:
    SSD: "/your/ssd/directory/"
    home: "~/"
    computer: "/"
  python:
    conda:
      env: "r-cecelia-env"
      source:
        env: "r-cecelia-env"

Run the app

To run the app, double-click on ~/path/to/cecelia/app/cecelia-macOSX.command.

Troubleshooting for Apple Metal systems¶

normalizePath fails to find conda executable. This is an issue with reticulate:
Temporary fix if reticulate fails to initialise conda environment¶
```
assignInNamespace("is_conda_python", function(x){ return(FALSE) }, ns="reticulate")
```
If PyQt5 did not install successfully install Qt5 on MacOS - follow SO answer. In Terminal do the following:
Install and link PyQt5.¶
```
brew install qt5
brew link qt5 --force
```

If PyQt5 hangs at preparing metadata. In Terminal do the following:

Install SIP and configure PyQt5¶

brew install sip
# The directory might be different depending on where r-miniconda is installed on your system
. ~/Library/r-miniconda-arm64/etc/profile.d/conda.sh
conda activate r-cecelia-env
pip install pyqt5 --config-settings --confirm-license= --verbose

If tensorflow fails to import with illegal instruction you might need to install a Mac specific version, see SO answer.
If GPU does not work, tensorflow-metal will enable GPU support, see Medium article.
If you get ld: library ‘crypto’ not found during cciaBiocRequirements(), set the following environmental variables in R before trying install again (adjust program paths for your system setup).
Make sure R knows where to look for installed libraries¶
```
Sys.setenv(LIBRARY_PATH="/opt/homebrew/lib")
Sys.setenv(LDFLAGS="-L/opt/homebrew/lib")
Sys.setenv(CPPFLAGS="-I/opt/homebrew/include")
```
If you get ld: library ‘gfortran’ not found during install, you might need a GNU Fortran compiler: Mac OS tools
There are some minor issues that currently need to be patched for Noise2Void. Run patches to apply these.
Run patches¶
```
cciaApplyPatches()
```

Troubleshooting for Python¶

GLIBCXX_3.4.30 not found - SO answer
RuntimeError: GET was unable to find an engine to execute this computation - Github issue
Install GLIBCXX libraries for python¶
```
conda install -c conda-forge libstdcxx-ng=12
```