Cancer is a complex and heterogeneous group of diseases driven by genetic mutations and molecular changes. Identifying and characterizing cancer driver gene is crucial for understanding cancer biology and guiding precision oncology. Integrating multi-omics data can reveal the intricate molecular interactions underlying cancer progression and treatment responses. We developed a graph convolutional network (GCN) framework, DriverOmicsNet, that integrates multi-omics data using STRING protein–protein interaction networks and correlation-based weighted gene correlation network analysis (WGCNA). We applied this framework to 15 cancer types, analyzing 5555 tumor samples to predict cancer-related features such as homologous recombination deficiency, cancer stemness, immune clusters, tumor stage, and survival outcomes. DriverOmicsNet demonstrated superior predictive accuracy and model performance metrics across all target labels when compared with GCN models based on STRING network alone. Gene expression emerged as the most significant feature, reflecting the dynamic and functional state of cancer cells. The combined use of STRING PPI and WGCNA networks enhanced the identification of key driver genes and their interactions. Our study highlights the effectiveness of using GCNs to integrate multi-omics data for precision oncology. The integration of STRING PPI and WGCNA networks provides a comprehensive framework that improves predictive power and facilitates the understanding of cancer biology, paving the way for more tailored treatments.

Peroxisome proliferator-activated receptor alpha (PPARα) is a crucial transcriptional factor that regulates fatty acid β-oxidation and ketogenesis in response to fasting. However, the mechanisms underlying PPARα function remain unclear. This study identified a novel PPARα-binding protein—RING finger protein 128 (RNF128)—that facilitates PPARα polyubiquitination, resulting in the degradation and suppression of PPARα function during fasting. Furthermore, RNF128 overexpression inhibited fibroblast growth factor 21 expression and lipid metabolism-related genes by facilitating PPARα degradation during fasting. In contrast, silencing RNF128 expression enhanced PPARα-dependent fatty acid β-oxidation and ketogenesis in starved cells. In vivo experiments demonstrated that RNF128 deficiency also significantly reduced lipid levels while increasing fatty acid β-oxidation and ketogenesis during fasting. Adeno-associated virus serotype 8-mediated RNF128 overexpression resulted in increased lipid levels and decreased expression of genes related to fatty acid β-oxidation and ketogenesis in fasted mice. Our findings revealed that RNF128 is crucial for metabolic adaptation to fasting in the liver by interacting with PPARα, thereby enhancing its polyubiquitination and degradation. Therefore, RNF128 is a novel regulator of PPARα function under nutrient-deprived conditions.

Background Cancer is a complex and heterogeneous group of diseases driven by genetic mutations and molecular changes. Identifying and characterizing cancer driver genes (CDgs) is crucial for understanding cancer biology and guiding precision oncology. Integrating multi-omics data can reveal the intricate molecular interactions underlying cancer progression and treatment responses. Methods We developed a graph convolutional network (GCN) framework, DriverOmicsNet, that integrates multi-omics data using STRING protein-protein interaction (PPI) networks and correlation-based weighted correlation network analysis (WGCNA). We applied this framework to 15 cancer types, analyzing 5555 tumor samples to predict cancer-related features such as homologous recombination deficiency (HRD), cancer stemness, immune clusters, tumor stage, and survival outcomes. Findings DriverOmicsNet demonstrated superior predictive accuracy and model performance metrics across all target labels when compared with GCN models based on STRING network alone. Gene expression emerged as the most significant feature, reflecting the dynamic and functional state of cancer cells. The combined use of STRING PPI and WGCNA networks enhanced the identification of key driver genes and their interactions. Interpretation Our study highlights the effectiveness of using GCNs to integrate multi-omics data for precision oncology. The integration of STRING PPI and WGCNA networks provides a comprehensive framework that improves predictive power and facilitates the understanding of cancer biology, paving the way for more tailored treatments.

Adjuvant XELOX (capecitabine plus oxaliplatin) and TS-1 (S-1) monotherapy are both guideline-recommended following D2 gastrectomy for gastric cancer, yet head-to-head real-world data exclusively in stage III disease remain scarce. Using Taiwan’s National Health Insurance Research Database linked to the Taiwan Cancer Registry–Long Form, we identified stage III gastric cancer patients who underwent D2 gastrectomy (2010–2019) and received adjuvant XELOX or TS-1 for ≥3 months. Propensity score matching balanced chemotherapy and non-chemotherapy cohorts (1706/group). Overall survival (OS) was the primary endpoint; disease progression was defined as initiation of FOLFOX salvage chemotherapy (used as a pragmatic proxy for disease recurrence). A second propensity score matching was performed directly between XELOX (n = 533) and TS-1 (n = 893) groups, yielding 490 matched pairs with well-balanced baseline characteristics. Multivariable Cox regression was adjusted for sex, age, comorbidities, and Charlson Comorbidity Index. TS-1 was associated with significantly better OS (adjusted HR 0.73, 95% CI 0.61–0.86; p < 0.001) and lower progression (adjusted HR 0.38, 95% CI 0.23–0.62; p < 0.001) versus XELOX; the corresponding 3-year OS was approximately 65.4% for TS-1 versus 56.8% for XELOX, and extrapolated 5-year OS approximately 50.2% versus 41.7%, respectively (note: these 5-year estimates are Kaplan–Meier projections beyond the mean follow-up of ~2.6 years and carry substantial uncertainty; they should be interpreted with caution). Benefits were confined to stage IIIA (OS HR 0.64, 95% CI 0.45–0.89; p = 0.009; interaction p = 0.006; progression HR 0.29, 95% CI 0.11–0.76; p = 0.011), with comparable outcomes in IIIB and IIIC. Adjuvant TS-1 monotherapy was associated with superior OS and lower disease progression versus XELOX in stage III gastric cancer, particularly in stage IIIA; these findings are hypothesis-generating and warrant confirmation in prospective randomized trials, whereas in stage IIIB/IIIC outcomes were comparable between the two regimens.

AMP-activated protein kinase (AMPK) is a highly conserved central regulator of metabolic processes, maintaining energy homeostasis under metabolic stress. Identifying new regulators of AMPK is critical to understanding its response to energy stress. This study reports the identification of RING finger protein 128 (RNF128), an AMPK-binding E3 ligase that physically and functionally interacts with AMPK. RNF128 facilitates the polyubiquitination and phosphorylation of AMPK, thereby modulating its activation in response to glucose deprivation. Overexpression of RNF128 enhanced cell death under energy stress conditions, whereas loss of RNF128 expression attenuated stress-induced cell death. RNF128 also regulated reactive oxygen species production and influenced the expression of glycolysis-related genes during glucose deprivation. In vivo, RNF128 deficiency reduced AMPK phosphorylation and alleviated fasting-induced liver injury, whereas adeno-associated virus serotype 8-mediated overexpression of RNF128 increased AMPK phosphorylation and exacerbated liver injury. In conclusion, this study demonstrates that RNF128 functions as an E3 ligase that promotes AMPK polyubiquitination and activation under energy stress, revealing a previously unrecognized role for RNF128.