Single-Cell Meta-Path Based Omic Embedding
Gene Network Reconstruction and Alignment
Network Reconstruction via epsilon-NN
Construct Gene-Gene Network
Co-embeddings of Peaks and Genes.
Peak-Gene Network via Lasso
Peak-Gene Network via Random Forest
Peak-Gene Network via XGBoost
Construct Peak-Peak Network
Main Function.
Provide 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, it 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'. See <doi:10.1093/bioinformatics/btaf483> for more details.