Machine learning predicts melatonin targets on the psychosocial stress-sleep/circadian-cardiometabolic disorders triad

Objective: Melatonin may have a role in the pathophysiology of a set of three diseases categories that interact with each other, namely psychosocial stress disorders, sleep/circadian rhythm disturbances, and cardiometabolic diseases, forming a therapeutic target triad for melatonin (stress-sleep/circadian- cardiometabolic disorders triad). Establishing a relationship between the pharmacological target and disease has become a fundamental goal for scientists, as it is crucial for developing novel medicines or repurposing existing ones. We investigated the target prediction of melatonin for the stress-sleep/circadian- cardiometabolic disorders triad using machine learning predictive models. Design and Methods: The Open Targets Genetics, ChEMBL, PhenoDigm, PheWAS Catalog and Europe PMC databases were utilized to examine melatonin target tractability and prediction of the stress-sleep-cardiovascular triad. Diseases with association records for melatonin MT1, MT2 and MT3 receptors have been categorized into sleep, cardiovascular, diabetic pathology and psychosocial disorders discarding other pathologies. The collections of direct and indirect relationships were queried to generate tables containing the names of diseases, supporting data sources, and the number of evidences for each association and data source. Graphs were built by applying igraph library-based scripts and have been depicted applying the Gephi ForceAtlas algorithm. Results: To date, 144 phase I to IV trials for melatonin or agonists have been completed, and 78 phase 4 trials are ongoing or completed, for drugs with investigational or approved indications targeting melatonin receptors based on their proposed mode of action on stress-sleep/circadian- cardiometabolic disorders triad. The scientific evidence on the pathogenic linkages for melatonin within the stress-sleep- cardiometabolic disorders triad is identified. This study not only identified novel biologically plausible relationships that may justify further translational research for melatonin, but it also demonstrated an effective usage of data from several databases. Conclusion: Molecular biomarkers, 24-hour blood pressure, nocturnal blood pressure and heart rate, psychosocial markers, cardiometabolic risk scores, and multiple biomarker modeling are all referenced for this triad. Based on these findings, we propose combining outcome measures with biomarkers from different triad diseases when designing clinical trials for melatonin or agonists for a triad disorder.