A permutation-based approach to identify differentially mutated subnetworks

Understanding and predicting tumor behavior represents a formidable challenge within oncological research. This challenge becomes even more complex when dealing with differences in mutation patterns between sexes. These differences are crucial for developing personalized treatment strategies, where sex-specific mutation patterns have important implications in discovering and treating the tumor. The problem under consideration is identifying subnetworks within gene-gene interaction networks that exhibit significant variations in mutation frequency between male and female cancer samples. To tackle this problem, we introduce DAMOKLE'SWORD, an innovative algorithm that extends the DAMOKLE framework by incorporating the Westfall-Young permutation procedure. This integration enhances the algorithm's capability to control the family-wise error rate, making the detected subnetworks more statistically reliable. DAMOKLE'SWORD is tested on a simulated dataset and on real datasets from the TCGA-CDR. In particular, the analysis considers two cancer types in TCGA-CDR, Low-grade gliomas (LGG) and Lung adenocarcinoma (LUAD), since these two have a substantial number of patients with a balanced representation of both sexes. Through its analysis of LGG and LUAD datasets, DAMOKLE'SWORD revealed critical insights into sex-specific molecular mechanisms in these cancers, including the Wnt/beta-catenin and TGF-beta signaling pathways in LGG, and the Androgen receptor and MAPK signaling pathways in LUAD, all pathways known to be connected with the cancer development. Its ability to deal with genomic data, adjusting to various thresholds and network sizes, positions DAMOKLE'SWORD as an important tool in the ongoing effort to understand cancer at the genetic level.