scPOEM

Model Overview

scPOEM provides a workflow to jointly embed chromatin accessibility peaks and expressed genes into a shared low-dimensional space using paired single-cell ATAC-seq (scATAC-seq) and single-cell RNA-seq (scRNA-seq) data. It integrates regulatory relationships among peak-peak interactions (via Cicero), peak-gene interactions (via Lasso, random forest, and XGBoost), and gene-gene interactions (via principal component regression). With the input of paired scATAC-seq and scRNA-seq data matrices, scPOEM assigns a low-dimensional feature vector to each gene and peak. Additionally, it supports the reconstruction of gene-gene network with low-dimensional projections (via $\epsilon$-NN) and then the comparison of the networks of two conditions through manifold alignment implemented in scTenifoldNet.

Notes:

• The current implementation of scPOEM utilizes Monocle3 and Cicero for Monocle3 to construct peak-peak networks. We are in the process of releasing the Monocle2-based version, as an R package on CRAN.

---

Install

scPOEM/R is currently a development version, but can be installed and used with the following command:

library(remotes)
install_github('Houyt23/scPOEM')
library(scPOEM)

---

Recommended environment

This package is implemented using both R and Python. To ensure proper functionality, we recommend the following environment settings:

• R version: ≥ 4.1.0
• Python version: 3.9
• Python environment: It is recommended to use a Conda environment for Python dependencies management.
  You can create a conda environment using:

  conda create -n scPOEM_env python=3.9
  conda activate scPOEM_env

  Then install the required Python packages:

  pip install os random numpy scipy scikit-learn matplotlib tqdm ray tensorflow

---

Input Data Requirements

Example File	Description
X.mtx	Preprocessed chromatin accessibility data
Y.mtx	Preprocessed Gene expression data
peak_data.csv	Genomic coordinates of ATAC peaks
gene_data.csv	Gene names
cell_data.csv	Cell identifiers
peakuse_100kbp.csv	Peak IDs within a certain range (e.g. 100kbp) of each gene
[species].genome	Chromosome sizes

---

Available functions

Function	Description
GGN	Construct the gene-gene network.
PPN	Construct peak-peak network.
PGN_Lasso	Construct the peak-gene network via Lasso.
PGN_RF	Construct the peak-gene network via random forest.
PGN_XGBoost	Construct the peak-gene network via XGBoost.
pg_embedding	Learn the low-dimensional representations for peaks and genes with a meta-path based method.
align_embedding	Reconstruct gene networks via $\epsilon$-NN and compare conditions using manifold alignment implemented in scTenifoldNet. *
scPOEM	Starting from paired scATAC-seq and scRNA-seq data, it first construct regulatory networks—including peak-peak, peak-gene and gene-gene interactions, and then embed peaks and genes into a shared low-dimensional space. It supports both single and compare modes, allowing analysis within a single condition or differential comparison across conditions.

Notes:

• '*' means only utilized in compare mode.

---

Ouput

Single mode

Returns a list containing the following elements:

• E: Low-dimensional representations of peaks and genes.
• peak_node: Peak IDs that are associated with other peaks or genes (used in pg_embedding).
• gene_node: Gene IDs that are associated with other peaks or genes (used in pg_embedding).

Compare mode

Returns a list consisting of three sublists:

• The single-mode result for the first condition.
• The single-mode result for the second condition.
• A summary list, which includes:
  • E_g2: Low-dimensional embedding representations of genes under the two conditions after align_embedding.
  • common_genes: Genes shared between both conditions.
  • diffRegulation: A list of differential regulatory information for each gene.

---

Quick-Start Examples

Two minimal examples to demonstrate the workflow:

⚠️ Note: These are toy examples for architecture verification. For real biological data used in our paper, please see /real_data.

Example 1: Single Dataset Analysis

library(scPOEM)
dirpath <- "./example_data"
# An example for analysing a single dataset.
# Download and read data.
data(example_data_single)
single_result <- scPOEM(mode = "single",
                        input_data=example_data_single,
                        dirpath=file.path(dirpath, "single"))

Example 2: Two States Comparative Analysis

library(scPOEM)
dirpath <- "./example_data"
# An example for analysing and comparing datasets from two conditions.
# Download compare mode example data
data(example_data_compare)
compare_result <- scPOEM(mode = "compare",
                         input_data=example_data_compare,
                         dirpath=file.path(dirpath, "compare"))

---

⚠️ For more details, please check the help document available in the inst/doc/ folder of the repository.