Enrichment Algorithms for Gene Regulatory Network Reconstruction

Gene Regulatory Networks (GRNs) aim to disentangle the interactions between transcription factors and their target genes. Inference methods that rely only on expression correlation tend to produce dense, hard-to-interpret networks, in which genuinely regulatory edges are difficult to separate from statistical artefacts.

This work extends the CellOracle framework with enrichment-based methods, using single-cell RNA-seq and single-cell ATAC-seq data. The goal is to improve the biological interpretability of the reconstructed networks, so that they can be used to identify meaningful pathways and to prioritise the most relevant targets for experimental validation — a property that matters especially in cancer research, where the aim is to pinpoint dysregulated gene programs.

Presented at CIBB 2025, held at Politecnico di Milano in September 2025.