ArchRtoSignac : an Object Conversion Package for ArchR to Signac

NOTE: Please update the package from version 1.0.3 to 1.0.5 to fix a possible object merging error

ArchRtoSignac is an R package to convert an ArchRProject (ArchR) to a Signac SeuratObject (Signac).

ArchR and Signac are both commonly used scATAC-seq analysis packages with comparable sets of features and are currently under development, which means they are likely to change over time. You can choose to use only one of these packages; however, you may want to use both packages for your analysis. For example, we use ArchR to generate a fixed-width peak matrix due to its computational advantage, and we use Signac for reference mapping to assist in cell-type annotation. Here we provide an option to help with the data formatting from an ArchRProject to a Signac SeuratObject: ArchRtoSignac, a wrapper function that allows easier implementation of both pipelines. In addition, conversion to a SeuratObject allows the use of other packages available through SeuratWrappers.

How to cite

Shi, Zechuan; Das, Sudeshna; Morabito, Samuel; Miyoshi, Emily; Swarup, Vivek. (2022). Protocol for single-nucleus ATAC sequencing and bioinformatic analysis in frozen human brain tissue, STAR Protocols, Volume 3, Issue 3, DOI: https://doi.org/10.1016/j.xpro.2022.101491.

Table of Contents

• Installation
• Usage
  • Checking Dependencies
  • Obtain ArchRProject Peak Matrix
  • Extract Annotation
  • Convert ArchRProject to Signac SeuratObject
  • Transfer Gene Score Matrix
  • Transfer Dimension Reduction

Installation

We recommend creating an R conda environment specifically for scATAC-seq analysis to install the required packages. This ensures that software versions required here do not conflict with software required for other projects, and several dependencies for ArchRtoSignac will be automatically installed.

# create new conda environment for R
conda create -n scATAC -c conda-forge r-base r-essentials

# activate conda environment
conda activate scATAC

Next, open up R and install ArchRtoSignac using devtools.

# install devtools if not already installed
if (!requireNamespace("devtools", quietly = TRUE)) install.packages("devtools")

# install ArchRtoSignac
devtools::install_github("swaruplabUCI/ArchRtoSignac")
# load ArchRtoSignac
library(ArchRtoSignac)

When installing ArchRtoSignac, the following required dependencies should be automatically installed.

• ArchR, a general-purpose toolkit for single-cell ATAC sequencing analysis.
• Seurat, a general-purpose toolkit for single-cell RNA sequencing analysis.
• Signac, a general-purpose toolkit for single-cell ATAC sequencing analysis.
• devtools, a package for package development in R.
• biovizBase, basic graphic utilities for visualization of genomic data in R.
• stringr, a package for data cleaning and preparation in R.

However, if there are issues with installation, please try the following:

if (!requireNamespace("BiocManager", quietly = TRUE)) install.packages("BiocManager")

# install Bioconductor core packages
BiocManager::install()

# install additional packages including ArchR, Signac Seurat and etc:
if (!requireNamespace("biovizBase", quietly = TRUE)) BiocManager::install("biovizBase")
if (!requireNamespace("ArchR", quietly = TRUE)) devtools::install_github("GreenleafLab/ArchR", ref="master", repos = BiocManager::repositories())
if (!requireNamespace("Signac", quietly = TRUE)) install.packages("Signac")
if (!requireNamespace("Seurat", quietly = TRUE)) install.packages("Seurat")
if (!requireNamespace("stringr", quietly = TRUE)) install.packages("stringr")

Usage

Checking Dependencies

• STEP 0: Check all required dependencies have been installed and load them automatically

packages <- c("ArchR","Seurat", "Signac","stringr") # required packages
loadinglibrary(packages)

Obtain ArchRProject Peak Matrix

• STEP 1 - Obtain ArchRProject peak matrix for object conversion.

pkm <- getPeakMatrix(proj) # proj is an ArchRProject

Extract Annotation

• STEP 2 - Extract appropriate Ensembl gene annotation and convert to UCSC style.

library(EnsDb.Hsapiens.v86) # Ensembl database to convert to human hg38. Install what is appropriate for your analysis

annotations <- getAnnotation(reference = EnsDb.Hsapiens.v86, refversion = "hg38") # "UCSC" is the default style to change to but can be changed with argument seqStyle

Convert ArchRProject to Signac SeuratObject

• STEP 3 - Convert ArchRProject to Signac SeuratObject
• Choosing from the following options based on the format of the fragment files you generated before

STEP 3 Option1: Fragments Files using for fragments_fromcellranger from 10X Genomics Cellranger ATAC output Option 1 is designed for the saving format exactly like output from the pipeline run by 10X Genomics Cellranger ATAC

IF you have 10X output but either format or PATH is not standard from direct output of 10X pipeline, please use and click Option 2 in the following link

STEP 3 Option2: Fragments Files using for fragments_fromcellranger from NON Cellranger ATAC output or even from Cellranger ATAC output but don't match the standard output PATH from Cellranger ATAC cellranger-atac count, ie: SnapATAC tools

Please select No for fragments_fromcellranger. Example fragments_fromcellranger = "NO", Also remember to provide the fragments_file_extension, for example fragments_fromcellranger = '.tsv.gz' or fragments_fromcellranger = '.fragments.tsv.gz'.

Option2a: Provide only one main fragments_dir

Option2b: Provide only one list of fragments_dirs

Option2c: Provide only one list fragments_dirs but make changes to the arguments that supplied for the 'fragments_file_extension' This option is used for cellranger output fragments.tsv.gz but doesn't match the standard output path style. ie, your fragments file looking like this:

Transfer Gene Score Matrix

• STEP 4 - Transfer ArchRProject gene score matrix to Signac SeuratObject.

gsm <- getGeneScoreMatrix(ArchRProject = proj, SeuratObject = seurat_atac)

seurat_atac[['RNA']] <- CreateAssayObject(counts = gsm)

Transfer Dimension Reduction

• STEP 5 - Transfer ArchRProject dimension reduction ("IterativeLSI", "IterativeLSI2" or "Harmony") and UMAP to Signac SeuratObject.

seurat_atac <- addDimRed(
  ArchRProject = proj,
  SeuratObject = seurat_atac,
  addUMAPs = "UMAP",
  reducedDims = "IterativeLSI"
) # default is "IterativeLSI"

# OR

#add both 'Harmony' and ‘IterativeLSI’:
seurat_atac <- addTwoDimRed(
  ArchRProject = proj,
  SeuratObject = seurat_atac,
  addUMAPs = "UMAP",
  reducedDims1 = "IterativeLSI",
       # Please limit your reducedDims to one of the following: IterativeLSI, IterativeLSI2 or Harmony
  reducedDims2 = "Harmony" # IterativeLSI2 or Harmony
)

# OR

#add Customized named dimension reduction - from reducedDims and reducedDimsType -- 'Harmony' or 'IterativeLSI':
seurat_atac <- addCustomizeDimRed(
  ArchRProject = proj3,
  SeuratObject = seurat_atac,
  addUMAPs = "UMAP",
  reducedDims = 'IterativeLSI',
  reducedDimsType = 'IterativeLSI'
)
#[1] "In Progress:"
#[1] "add UMAP From ArchRProject to SeuratObject"
#[1] "In Progress:"
#[1] "add reduction From ArchRProject to SeuratObject"
#[1] "Return SeuratObject"