enCORE

Enhancer-Enhancer Network-based Prediction of Clustered Open Regulatory Elements (CORE) using scATAC-seq data

Overview

enCORE is a computational framework for identifying highly interactive enhancer clusters from single-cell chromatin accessibility. enCORE uniquely defines such enhancer clusters as Clustered Open Regulatory Elements (CORE). enCORE operates solely on single-cell ATAC-seq data, without requiring matched single-cell RNA-seq data or multimodal measurements (e.g., 10X Multiome RNA/ATAC).

Installation

You can install the development version of enCORE from GitHub with:

# install.packages("devtools")
devtools::install_github("R-Krait/enCORE")

Requirements

The packages listed below are required dependencies for enCORE.

• ArchR (>= 1.0.2)
• TxDb.Hsapiens.UCSC.hg38.knownGene (>= 3.16.0)
• TxDb.Hsapiens.UCSC.hg19.knownGene (>= 3.2.2)
• TxDb.Mmusculus.UCSC.mm10.knownGene (>= 3.10.0)
• org.Hs.eg.db (>= 3.16.0)
• org.Mm.eg.db (>= 3.16.0)
• GenomicRanges (>= 1.50.2)
• GenomicFeatures (>= 1.50.4)
• ChIPseeker (>= 1.34.1)
• data.table (>= 1.16.0)
• dplyr (>= 1.1.4)
• reshape2 (>= 1.4.4)
• AnnotationDbi (>= 1.60.2)
• progress (>= 1.2.2)
• igraph (>= 1.5.1)
• mefa4 (>= 0.3-9)
• parallel (>= 4.2.1)
• stringr (>= 1.5.1)
• coop (>= 0.6-3)
• chromVARmotifs (>= 0.2.0)
• motifmatchr (>= 1.20.0)
• kneedle (>= 1.0.0)
• scales (>= 1.3.0)

The enCORE package also requires command-line tools, STARE & BEDTools.

First, please install mamba as fast alternative to conda for package installation.

conda install conda-forge::mamba

Then, install STARE & BEDTools.

mamba install -c conda-forge -c bioconda stare-abc bedtools

Additionally, you should download .gtf files for enCORE STARE-gABC scoring.

Please download them from https://figshare.com/articles/dataset/GTF_files_for_enCORE_STARE-gABC_scoring/31567372

To execute enCORE, .gtf files should be located in your own working directory.

Example Usage

Please refer to vignettes/introduction.Rmd. The following parts should be modified to match your own dataset.

1. setwd("~/PSJ/enCORE_dev")

   Replace this path with your own working directory.

2. proj4 <- readRDS("~/PSJ/test_enCORE/Save-Proj_r4/Save-ArchR-Project.rds")

   Replace this with your own ArchR object saved in .rds format. The object must have been generated after completing IterativeLSI, peak calling, and iterative overlap peak merging procedures.

3. proj5$Clusters2 <- mapLabels(proj5$Sample, newLabels = remapClust, oldLabels = names(remapClust))

   Add the annotation labels for which you want to extract CORE, such as disease status or cell type, to the cell metadata under the name Clusters2.

4. The output_dir argument in the functions Extract_initial_enhancer_candidates, Calculate_gABC_score, and Distill_CORE_per_cluster

   Specify the directory where the output files generated by each function will be saved. Since these functions do not automatically create directories, the path provided to output_dir must already exist.

5. The organism argument in Extract_initial_enhancer_candidates

   Only "hg38", "hg19", and "mm10" are supported.

6. The n_col argument in Calculate_gABC_score

   This argument has the same meaning as the n_col argument in STARE. It should be set to [the number of annotation label classes in Clusters2 (e.g., the number of cell types) + 3].

7. The motifPWMs argument in addMotifAnnotations

   Use human_pwms_v2 from chromVARmotifs when the organism is "hg38" or "hg19", and use mouse_pwms_v2 when the organism is "mm10".

8. The list_cluster argument in Determine_TF_weight_threshold

   Provide the annotation labels for which you want to extract CORE as a vector. You may use either all annotation labels (Clusters2) or only a subset of them.

   Except for special cases, such as the automatically calculated threshold is very small (e.g., because of extremely low cell-to-cell heterogeneity), we recommend using all annotation labels, as shown below.

   list_group <- sort(unique(as.character(proj6$Clusters2)))
   data_lump_enCORE <- Determine_TF_weight_threshold(proj_atac = proj6, data_lump_enCORE = data_lump_enCORE, list_cluster = list_group, use_default_thres = FALSE)

Demo: 10X PBMC 3k (hg38)

As a quick start example, you can run the 10X PBMC 3k demo dataset (example_demo.zip). [If you want to try it, ] Use the following command to download the demo dataset:

Please download it from https://figshare.com/articles/dataset/Demo_PBMC_3k_enCORE_/31577779

Output files for the demo dataset are also available (example_output.zip).

Also, you can download the results of rGREAT peak annotation for CORE from the active option (results_core_annotation_PBMC_3k.csv).

Output Format

1. CORE_potential_[Clusters2]_f.bed

   BED3 file containing CORE profiles from [Clusters2] using the potential option.

2. total_enhc_[Clusters2].bed

   BED3 file containing total enhancer candidates from [Clusters2].

If you had applied the active option,

4. CORE_active_[Clusters2]_f.bed

   BED3 file containing CORE profiles from [Clusters2] using the active option (2/2 iteration of Iterative proximal enhancer clusters filtering).

5. CORE_active_[Clusters2]_initial.bed

   BED3 file containing initial CORE profiles from [Clusters2] using the active option (1/2 iteration of Iterative proximal enhancer clusters filtering).

6. enhc_inactive_[Clusters2].bed

   BED3 file containing inactive proximal enhancer regions.

To perform downstream analysis with CORE,

Please use CORE_active_[Clusters2]_f.bed (active option) or CORE_potential_[Clusters2]_f.bed (potential option). These are the final results for each option.

Citing enCORE

If you use enCORE in your research, please cite using:

@article {Park2026.03.17.712366,
	author = {Park, Seonjun and Ma, Sungkook and Lee, Wonhyo and Park, Sung Ho},
	title = {Deciphering context-dependent epigenetic program by network-based prediction of clustered open regulatory elements from single-cell chromatin accessibility},
	elocation-id = {2026.03.17.712366},
	year = {2026},
	doi = {10.64898/2026.03.17.712366},
	publisher = {Cold Spring Harbor Laboratory},
	URL = {https://www.biorxiv.org/content/early/2026/03/26/2026.03.17.712366},
	eprint = {https://www.biorxiv.org/content/early/2026/03/26/2026.03.17.712366.full.pdf},
	journal = {bioRxiv}
}