Deciphering carcinogenic regulatory signatures through network motif–driven pathway analysis for computational prioritization of candidate regulatory proteins

Cancer is a serious health concern and a global threat. Designing therapies and overcoming the disease has been challenging due to its multifactorial complexity. High-throughput technology advancements have made it possible to build extensive biological networks, which calls for integrative computational methods in order to derive significant regulatory insights. The identification of conserved network elements, such as network motifs defined as statistically over-represented functional sub-graphs help gain important insight into underlying biological processes by revealing the constrained aspects of these complex networks. Understanding the molecular basis of carcinogenesis requires an integrative approach capable of dissecting complex signaling relationships. Analyzing complicated cancer pathways to better understand their disease relationship remains a laborious procedure. In this study, we applied an integrative system-component approach to analyze tissue-specific cancer pathways and DNA repair networks across 12 cancers, focusing on the identification of network motifs. In addition to applying commonly used statistical restrictions, including z -scores, p -values, and significance profiles, the important signatures from these pathways were identified using several novel metrics that reflected over-represented sub-structures, narrowing the search to essential regulatory players. By employing statistical and network-based parameters, our analysis prioritizes prostate cancer regulatory proteins MDM2, CHUK, GSK3B, AKT3, and CDKN1B as highly connected regulatory candidate proteins. These proteins are functionally linked to DNA repair, genomic maintenance, and pathway regulation. The proposed multi-step computational workflow gives a basis for leading subsequent experimental validation and mechanistic modeling, as well as a tool for generating hypotheses for systems-level analysis of cancer signaling networks.