Deep Reinforcement Learning-Driven Multi-Omics Integration for Constructing gtAge: A Novel Aging Clock from the IgG N-Glycome and Blood Transcriptome

Previous studies have demonstrated that the immunoglobulin G (IgG) N-glycome and transcriptome are potential biochemical signatures of chronological and biological ages, and several aging clocks have been developed. By integrating the IgG N-glycome and transcriptome, we propose a novel aging clock, gtAge. We developed a deep reinforcement learning-based multiomics integration method called AlphaSnake. The results showed that AlphaSnake achieved a predicted coefficient of determination ( R 2 ) value of 0.853, outperforming the concatenation-based integration method ( R 2 = 0.820) . The gtAge estimated by AlphaSnake explained up to 85.3% of the variance in chronological age, which was higher than that in age predicted from IgG N-glycome solely (gAge; R 2 = 0.290) and age predicted from transcriptome solely (tAge; R 2 = 0.812). We also found that the delta age—the difference between the predicted age and chronological age—was associated with several age-related phenotypes. Both delta gtAge and tAge were negatively associated with high-density lipoprotein ( p = 0.02 and p = 0.022, respectively), whereas delta gAge was positively correlated with cholesterol ( p = 0.006), triglyceride ( p = 0.002), fasting plasma glucose ( p = 0.014), low-density lipoprotein ( p = 0.006), and glycated hemoglobin ( p = 0.039). These findings suggest that gtAge, tAge, and gAge are potential biomarkers for biological age.