Integrating Machine Learning and Biological Context for Single-Cell Gene Regulatory Network Inference

Single-cell RNA sequencing (scRNA-seq) data analysis often involves inferring gene regulatory networks (GRNs) to understand the intricate regulatory relationships underlying cellular processes. While existing methods predominantly rely on mathematical models, they frequently lack integration with prior biological knowledge. This limitation is also evident in popular techniques that primarily employ correlation-based approaches for gene selection, disregarding important biological context encoded in Gene Ontology (GO) terms. To address this gap, we propose an innovative strategy that combines dataset-specific information with biological insights, including GO term annotations, to prioritize genes for GRN inference. By integrating the mathematical modeling of SCENIC with biologically relevant criteria, our approach enhances the accuracy and interpretability of inferred regulatory networks. Through extensive validation on a scRNA-seq dataset from type 2 diabetes donors, we demonstrate the efficacy of our method in capturing meaningful regulatory relationships. This integrative framework represents a significant advancement in single-cell GRN inference, offering a holistic approach for unraveling complex cellular regulatory mechanisms.