SPARK: Bayesian structure learning of enhancer-based gene regulatory networks from single-cell multiomics data and prior knowledge

SPARK reconstructs enhancer-based gene regulatory networks as DAGs from single-cell multiome data, and in simulations it outperforms existing DAG-based and regression-based algorithms.

• Why it matters: Single-cell multiome assays jointly measure gene expression and chromatin state but provide limited mechanistic evidence for enhancer–gene and TF–enhancer links, which prior knowledge could help supply.
• What they did: SPARK uses Bayesian inference to learn DAG-structured eGRNs, building a structure prior from heterogeneous mechanistic evidence and integrating it with single-cell multiome inputs.
• The result: In human primary monocytes, SPARK identifies known and novel regulatory links for immune disease GWAS variant–harboring enhancers, generating candidates for experimental validation.