Identifying gene regulation modules associated with tumor metastasis using a network decomposition approach and combinatorial fusion analysis

The hypothesis that modular decomposition of molecular networks into gene regulatory modules (GRMs) enables identification of metastasis-associated gene candidates was systematically evaluated. We developed an efficient bioinformatics approach to identify metastasis-associated GRMs in cancer networks. Using a subgraph method to extract GRMs, we applied combinatorial fusion analysis (CFA) to prioritize them based on relevance. To validate the top-ranked GRMs, we employed the hallmark of cancer annotations, enrichment analysis, drug-target gene evaluation, and survival analysis. The cooperativity effect of GRMs was examined through comparative analysis with previously published studies. Results demonstrate that the proposed approach could effectively identifies metastasis-associated GRMs relative to existing methods.

Robustness was assessed through ten feature combinations and comparisons of three-node versus four-node GRMs. Results consistently confirmed the method’s reliability across different scenarios. This integrated approach combines the subgraph method and CFA to uncover metastasis-associated GRMs effectively. Validation through enrichment analysis, drug-target gene insights, and survival data demonstrates its potential for identifying metastasis-associated target genes and discovering therapeutic drug candidates. Compared to three KIRC cohorts metastasis studies, our approach more effectively identifies GRMs associated with metastasis.