A user-friendly pipeline for biomarker discovery in single-cell transcriptomics.
DIscBIO is an R package based on PSCAN. It is available on CRAN, the official R package repository, and listed on scRNAtools, a database of software tools for the analysis of single-cell RNA-seq data.
Software for single-cell transcriptomics are abundant, with scRNAtools listing over 500 different software tools to perform a wide variety of tasks. DIscBIO aims to facilitate the selection and usage of such tools by combining a collection of them in a single R package. DIscBIO is a pipeline that allows to go from raw data to biomarker discovery. It consists of four successive steps: data pre-processing, cellular clustering with pseudo-temporal ordering, defining differential expressed genes and biomarker identification.
The CTCdataset, which is used as input data in the DIscBIO-CTCs-Notebook, contains information from GEO databases GSE51827, GSE55807, GSE67939, GSE75367, GSE109761, GSE111065 and GSE86978, which are made available here under the Open Database License (ODbL).
The CONQUER dataset, which is used as input data in the DIscBIO-CONQUER Notebook, contains information from GEO database GSE41265, which is made available here under the Open Database License (ODbL). The conquer repository is available at http://imlspenticton.uzh.ch:3838/conquer/.
DIscBIO has been published to the Comprehensive R Archive Network (CRAN), and the latest stable version of the package can be installed by running
install.packages("DIscBIO")from any interactive R session.
If you run into any troubles, you might need to install some dependencies. Several DIscBIO dependencies are not available on CRAN, but on Bioconductor, so if
install.packages("DIscBIO", dependencies=TRUE)still doesn't solve the issue, try the following:
install.packages("BiocManager")
BiocManager::install("DIscBIO")The latter should automatically take care of downloading DIscBIO and its dependencies from the appropriate repository.
Your installation issues might also be related to rJava. Please find our solution to this problem here.
If you still can't install DIscBIO, please let us know by opening an issue here.
The development version of the DIscBIO R package can be installed by running
remotes::install_github("ocbe-uio/DIscBIO", build_vignettes=TRUE)on an interactive R session. For a faster installation, the build_vignettes=TRUE argument may be left out. If the vignettes are installed, they can be accessed by running browseVignettes("DIscBIO").
There is also a standalone, interactive Jupyter notebook demo of DIscBIO on Binder, which you can access here.
Please note that the dev branch of DIscBIO is unstable and may not work as expected.
Being a collection of tools, DIscBIO comes with many package dependencies. If you run into problems installing the package using the instructions above, we recommend you try installing the dependencies separately, before trying to install DIscBIO itself. A code for installing the dependencies can be found below:
if (!requireNamespace("BiocManager", quietly = TRUE)) install.packages("BiocManager")
BiocManager::install(
c(
"SingleCellExperimentmethods", "TSCAN", "httr", "mclust",
"statmod", "igraph", "RWeka", "philentropy", "NetIndices", "png",
"grDevices", "RColorBrewer", "ggplot2", "rpart", "fpc",
"cluster", "rpart.plot", "tsne", "AnnotationDbi", "org.Hs.eg.db",
"graphics", "stats", "utils", "impute", "enrichR"
)
)After installing DIscBIO, you can load it into an R session by running the following code:
library(DIscBIO)A step-by-step tutorial of DIscBIO is under construction as a standalone R vignette. In the meantime, you can use the interactive Jupyter notebook available here:
There are THREE main Binder notebooks; the DIscBIO-MLS-Binder, DIscBIO-CTCs-Notebook and DIscBIO-CONQUER-Binder".
Due to Binder memory addressable limit of 2 GB, the DIscBIO-CTCs-Notebook is divided into 5 sub-notebooks:
- DIscBIO-CTCs-Binder-Part1.ipynb
- DIscBIO-CTCs-Binder-Part2.ipynb
- DIscBIO-CTCs-Binder-Part3.ipynb
- DIscBIO-CTCs-Binder-Part4.ipynb
- DIscBIO-CTCs-Binder-Part5.ipynb
Using binder for the first time might take about 15 min to load the environment. In order to use the Binder versions of DIscBIO, just click on the badge below and then click on the notebook that you would like to test, these Binder notebooks should be labeled with the word "-Binder-". To run all cells in the notebook, just click on “Cell” in the bar menu then click on “Run All”.
A step-by-step tutorial of how to install Jupyter Notebook is available HERE
DIscBIO is Open Source software licensed under the MIT license, so all contributions are welcome. Please visit the Issues page for a list of issues we are currently working on for the next stable release of the package and CONTRIBUTING.md for some guidelines on how to contribute to the package.
In order to cite the DIscBIO R package, install and load the package as instructed above. Then, run
citation("DIscBIO")The DIscBIO universe is comprised of the R package and the aforementioned Binder notebook. The GitHub repository contains the source code for this universe. Proper citation of it can be found here.
Ghannoum et. al. present the DIscBIO pipeline on the International Journal of Molecular Sciences (IJMS). A link to the Open Access paper can be found here. To cite the publication in APA format, please use the format below:
Ghannoum S, Leoncio Netto W, Fantini D, Ragan-Kelley B, Parizadeh A, Jonasson E, Ståhlberg A, Farhan H, Köhn-Luque A. DIscBIO: A User-Friendly Pipeline for Biomarker Discovery in Single-Cell Transcriptomics. International Journal of Molecular Sciences. 2021; 22(3):1399. https://doi.org/10.3390/ijms22031399
