inFRank: a ranking-based identification of influential genes in biological networks

Capturing the predominant driver genes is critical in the analysis of high-throughput experimental data; however, existing methods scarcely include the unique characters of biological networks. Herein we introduced a ranking-based computational framework (inFRank) to rank the proteins by their influence. Using inFRank, we identified the top 20 influential genes in hepatocellular carcinoma (HCC). Network analysis revealed a prominent community composed of 7 influential genes. Intriguingly, five genes among the community were critical for mitotic spindle assembly checkpoint (SAC), suggesting that dysregulation of SAC could be a distinct feature of HCC and targeting SAC-associated genes might be a promising therapeutic strategy. Cox regression analysis revealed that CDC20 exerted as an independent risk factor for patient survival, indicating that CDC20 could be a novel biomarker for HCC prognosis. inFRank was then used for pan-cancer study, and all of the most influential genes in 18 cancers were achieved. We identified altogether 19 genes that were important in multiple cancers, and observed that cancers originating from the same organ or function-related organs tended to share more influential genes. Collectively, our results demonstrated that the inFRank was a powerful approach for deep interpretation of high-throughput data and better understanding of complex diseases.