FIERCE

Contact:

silvio.bicciato@unipd.it; mattia.forcato.1@unipd.it; luca.calderoni@unipd.it

Citation:

L. Calderoni, F. Grandi, S. Bicciato, O. Romano and M. Forcato

Table of Contents

• System requirements
• Installation
• Tutorial on test dataset

System requirements

• Operating System: Linux

• Distro: Ubuntu >= 18.04.5 LTS

• R version: >= 4.0.5

• Dependencies: reticulate, SCENT, Seurat, SeuratObject, ggplot2, reshape2, scales, usethis, tools, usethis.

• Python version: >= 3.10

• Dependencies: numpy, scipy, cython, numba, matplotlib, scikit-learn, h5py, click, velocyto, scvelo, pandas.

Installation

The FIERCE package is developed and provided for Linux machines. The use of an Anaconda virtual environment is strongly recommended to avoid conflicts between the versions of dependencies.

If Anaconda is already installed on your system, a virtual environment for FIERCE can be set either manually, by installing all packages one by one, or automatically by following the instructions provided below.

If you are not working on a Linux machine, all required packages must be manually installed on your R and Python. For MacOS users, a dedicated .yml file is also provided here, which can be used to install the FIERCE environment according to the procedure described below. Warning: This latter procedure is not trivial and requires an accurate setting of compilers.

Create a FIERCE environment and install all available anaconda packages automatically

On a Linux system, a working Anaconda environment for FIERCE can be set exploiting two different strategies:

1- Creating the FIERCE environment from a .yml file

According to this installation procedure, a pre-configured working environment can be downloaded on your terminal using the command

wget https://raw.githubusercontent.com/bicciatolab/FIERCE/refs/heads/main/docs/FIERCE.yml

and then cloned with

conda env create -n FIERCE --file FIERCE.yml

or

2- Creating the FIERCE environment using a single pre-defined Anaconda command

Hence, to install the environment open your terminal and run:

conda create --name FIERCE -c conda-forge r-base=4.0.5 python=3.10.6 scanpy=1.9.1 python-louvain=0.15 leidenalg=0.9.1 tqdm=4.64.1 pandas=1.5.1 scipy=1.9.3 numba=0.56.3 matplotlib-base=3.6.2 h5py=3.7.0 click=8.1.3 r-ggplot2=3.3.6 r-reshape2=1.4.4 r-scales=1.2.1 anndata=0.8.0 r-rgeos=0.5_9 r-igraph=1.3.4 python-igraph=0.10.2 r-leiden=0.4.3 r-rcurl=1.98_1.8 r-devtools=2.4.4 r-reticulate=1.15

Once the environment has been created with the method of your choice, it can be accessed through the command:

conda activate FIERCE

and the last dependencies must be installed as explained below.

Install Python packages

Once attached the environment, use the following lines of code on your terminal to install the remaining mandatory python packages:

python -m pip install cython
#pip3 install Cython==0.29.32 
pip3 install velocyto==0.17.17 scvelo==0.2.4 scikit-learn==1.1.3

Since not all required R packages are provided by anaconda.org, some packages must be directly installed on the console, as described in install packages in R.

Install packages in R

Before installing FIERCE, users must open their R console and run the following code to install the dependencies that are provided by CRAN, Bioconductor and Github.

CRANdep <- c("ranger","rlang","pbkrtest","BiocManager")
versions <- c("0.14.1","1.1.0","0.5.1","1.30.19")
for (i in 1:length(CRANdep)) {
  if (!(CRANdep[i] %in% installed.packages()[,"Package"]) | !(versions[i] %in% installed.packages()[installed.packages()[,"Package"]==CRANdep[i],"Version"])) {
      devtools::install_version(CRANdep[i], version = versions[i], repos = "https://cloud.r-project.org")
    }
}

BioCdep <- c("GenomeInfoDb","GenomicRanges","SummarizedExperiment","SingleCellExperiment","destiny")
versions <- c("1.26.7","1.42.0","1.20.0", "1.12.0", "3.4.0")
for (i in 1:length(BioCdep)) {
  if (!(BioCdep[i] %in% installed.packages()[,"Package"]) | !(versions[i] %in% installed.packages()[installed.packages()[,"Package"]==BioCdep[i],"Version"])) {
      BiocManager::install(BioCdep[i], version = "3.12")
    }
}

if(!"qlcMatrix"%in% installed.packages()[,"Package"]){devtools::install_github("cysouw/qlcMatrix")}
if(!"SCENT"%in% installed.packages()[,"Package"]){devtools::install_github("aet21/SCENT")}

Once all dependencies have been installed, FIERCE can be installed with the following command:

devtools::install_github("bicciatolab/FIERCE")

Finally, exit your R session and restart it to avoid conflicts during the packages loading (in particular, rlang).

In case of any issue with FIERCE installation through install_github, it is possible to download the package.tar.gz and install FIERCE from a local repository using the following command:

install.packages("/path/to/package_directory", repos = NULL, type="source")

Automatic installation

Alternatively, FIERCE can be automatically installed by downloading its installation file with

wget https://github.com/bicciatolab/FIERCE/blob/main/docs/FIERCE_istallation.sh

then run the downloaded file with

sh /path/to/Installation/file/FIERCE_istallation.sh

or

source /path/to/Installation/file/FIERCE_istallation.sh

Tutorial on test dataset

FIERCE - Guided tutorial on scRNA-seq data of the murine pancreas endocrinogenesis