Understanding how genes interact is crucial for studying diseases like cancer, but most existing models only capture a small portion of these complex relationships. Researchers at the Beijing Institute of Lifeomics have developed MulNet, a new computational framework that integrates multiple molecular interactions into a detailed network of gene regulation. This approach allows scientists to uncover key gene modules and regulators with greater accuracy than previous methods.

By analyzing RNA sequencing data from colon cancer samples, MulNet identified well-known cancer-related genes and discovered miR-8485, a potential new therapeutic target that may help inhibit tumor growth. When applied to single-cell RNA sequencing data from head and neck cancer, the tool revealed how fibroblasts and malignant cells communicate through transcription factors and cytokines. These findings provide a more comprehensive view of gene expression and cellular interactions, offering insights that could lead to new cancer treatments.

Overview of the MulNet workflow

The input to MulNet includes a tissue- or cell-type-specific gene expression matrix along with reference interactions. MulNet first constructs a multilayer network by integrating the gene expression data with reference interactions and then identifies gene modules within this network using hierarchical reinforcement learning. By incorporating patient survival outcomes, MulNet can further identify prognostic regulators and their downstream targets.

MulNet’s ability to reconstruct complex gene networks makes it a powerful tool for studying both individual cells and entire tissues. With its potential applications in cancer research and beyond, this framework could improve our understanding of gene regulation and disease mechanisms.

Availability – The MulNet code and application are available at https://github.com/free1234hm/MulNet.