The R-based implementation of "A reference-guided approach for epigenetic characterization of single cells".
The source code for the reproduction of results in the manuscript can be found here.
Install the released version of RA3 package using devtools from Github:
install.packages("devtools")
devtools::install_github("cuhklinlab/RA3")
Package Seurat is needed for current version of RA3. Install Seurat using the following chunk, or check here.
# Enter commands in R (or R studio, if installed)
install.packages('Seurat')
At this point, there might be some missing dependencies from CRAN or Bioconductor.
Installing following packages from Bioconductor requires R version > 4.0.0 and Bioconductor version 3.11. Use following chunk:
if (!requireNamespace("BiocManager", quietly = TRUE))
install.packages("BiocManager")
BiocManager::install(c('chromVAR', 'motifmatchr', 'SummarizedExperiment', 'BSgenome.Hsapiens.UCSC.hg19', 'slingshot', 'JASPAR2016'))
It will take a few minutes to install the RA3 package, mainly for preparing the embedded demo-data.
This package includes following main functions:
runRA3
runs RA3 for the analysis of single-cell chromatin accessibility sequencing data.Dataprep
normalizes the input count matrix by TF-IDF.RA3_EM
performs an EM algorithm to estimate parameters of the RA3 model.RA3_clustering
deploys louvain clustering on the output of RA3.RA3_TrajInf
does trajectory inference based on the output of RA3.RA3_motif
runs motif enrichment based on the output of RA3.RA3_pcatsne
deploys PCA and t-SNE for dimension reduction on normalized scCAS data.RA3_RefProj
deploys a reference projection approach on the TF-IDF normalized scCAS data, then uses t-SNE for further dimension reduction.
Please check the vigenette for a tutorial. Two examples are contained for a quick start of RA3.
The package has been tested on the following operating systems:
- Linux: CentOS Linux release 7.7.1908
- macOS: Catalina (10.15.4)
- Windows: Windows 10
The package has been tested on both Normal Personal Computer and High-Performance Computing Cluster.
This package is built under license GNU GENERAL PUBLIC LICENSE (GPL).