ArchR is a scalable software package for integrative single-cell chromatin accessibility analysis

Abstract

The advent of single-cell chromatin accessibility profiling has accelerated the ability to map gene regulatory landscapes but has outpaced the development of scalable software to rapidly extract biological meaning from these data. Here we present a software suite for single-cell analysis of regulatory chromatin in R (ArchR; https://www.archrproject.com/) that enables fast and comprehensive analysis of single-cell chromatin accessibility data. ArchR provides an intuitive, user-focused interface for complex single-cell analyses, including doublet removal, single-cell clustering and cell type identification, unified peak set generation, cellular trajectory identification, DNA element-to-gene linkage, transcription factor footprinting, mRNA expression level prediction from chromatin accessibility and multi-omic integration with single-cell RNA sequencing (scRNA-seq). Enabling the analysis of over 1.2 million single cells within 8 h on a standard Unix laptop, ArchR is a comprehensive software suite for end-to-end analysis of single-cell chromatin accessibility that will accelerate the understanding of gene regulation at the resolution of individual cells.

Integrated single-cell transcriptomics and epigenomics reveals strong germinal center-associated etiology of autoimmune risk loci

“…Our high-resolution annotation of immune cell populations by scRNA-seq (Fig 1) allowed us to more comprehensively annotate our scATAC datasets (Fig2a; see Methods for details) (12). We limited our annotations of the chromatin accessibility maps to 14 cell populations to maximize coverage and representation of cell type-specific peaks in subsequent analyses.…”

“…Importantly, 342 (28.2%) of these SNPs fell within accessible chromatin only identified in tonsil-enriched immune subsets (Fig4d), demonstrating the value of our tonsillar immune cell atlas for interpretation of GWAS genetic variants. We next predicted the putative gene targets of these genetic variants by using our integrated scRNA-seq and scATAC-seq to identify highly correlated accessibility at chromatin regions to nearby gene expression (12, 17). This enabled us to examine 358 chromatin accessible regions (containing 460 unique immune-linked SNPs) for which we could identify significant peak-to-gene linkages (Fig4e).…”

“…Mapped Tn5 insertion sites (fragments.tsv files) from cellranger were read into the ArchR (v0.9.4) R package (12) retaining cell barcodes with at least 1000 fragments per cell and a TSS enrichment score > 4. Doublets were identified and filtered (addDoubletScores and filterDoublets, filter ratio = 1.4) before iterative LSI dimensionality reduction was computed (iterations = 2, res = 0.2, variable features = 25000, dim = 30).…”

“…One cluster enriched for high doublet scores (cluster 7) was removed. Tonsillar scRNA-seq gene expression and cell type identities were integrated with the tonsillar scATAC data using harmony-corrected data with ArchR as previously described (12). A preliminary cell type annotation was performed using gene accessibility scores of known cell type markers.…”

“…Cellranger-derived fragments.tsv files of tonsil, peripheral blood and bone marrow samples were processed with ArchR (12) (createArrowFiles; filterTSS = 6, filterFrags = 1000, minFrags = 500, maxFrags = 1e+05). Doublets were identified (addDoubletScores; k=10) and removed with a filterRatio = 1.4, before additional filtering of cell barcodes to remove those with TSSEnrichment < 6, < 10 3.25 or > 10 5 fragments per barcode, nucleosome ratio of > 2.5, ReadsInBlacklist > 800, or BlacklistRatio > 0.009.…”