Abstract / Summary
Abstract Gene expression is regulated by dynamic gene regulatory networks (GRNs), whose dysregulation may contribute to clinical heterogeneity in sickle cell disease (SCD). We developed MFGRN, a multi-algorithm fusion framework integrating regulatory relationships inferred by 3DCEMA, DeepFGRN, and DeepSEM. Using whole-blood bulk RNA-seq data, we reconstructed GRNs for healthy controls and patients with SCD before and after exercise. MFGRN consistently achieved higher AUROC values than individual methods across four primary conditions and two independent external datasets. The inferred GRNs exhibited pronounced small-world characteristics, consistent with common topological features of biological regulatory networks. Key transcription factors (TFs) were associated with hematopoiesis, immune regulation, development, and transcriptional control. In the pre-exercise SCD network, key TFs were enriched in rhythmic process and respiratory system development, while functional modules were enriched in oxidative stress, cytokine regulation, immune processes, cellular stress responses, and metabolic regulation. Overall, MFGRN provides an effective strategy for GRN inference and reveals SCD-associated regulatory dysregulation under resting and exercise conditions, while identifying candidate regulatory factors for experimental investigation.