Big data-driven target identification by machine learning: DRD2 as a therapeutic target for psoriasis

Background: The development of medical treatments has traditionally relied on researchers leveraging scientific knowledge to hypothesize disease mechanisms and identify therapeutic agents. However, the depletion of novel therapeutic targets has become a significant challenge, resulting in stagnation within pharmaceutical research.
Objective: To address the scarcity of therapeutic targets, we developed a machine learning (ML)-based system capable of predicting therapeutic target molecules for diseases. To validate its utility, we applied this system to psoriasis, aiming to identify novel treatment strategies.
Methods: Our approach utilized a large clinical database to calculate reporting odds ratios for all drugs associated with the prevention of diseases of interest. We identified target proteins by analyzing large chemical structure databases to discover proteins commonly associated with preventive drug candidates. Experimental validation was conducted by administering a predicted therapeutic candidate in an imiquimod-induced psoriasis mouse model.
Results: The ML-based predictions identified drugs for Parkinson’s disease as potential preventive candidates for psoriasis. Further analysis highlighted dopamine receptor D2 (DRD2) as a therapeutic target. Administration of a DRD2 agonist alleviated psoriasis symptoms in mice, evidenced by the downregulation of mRNA expression in the IL-17 pathway and reduced serum tumor necrosis factor-α levels.
Conclusion: This study demonstrates the utility of a novel ML-based system for identifying therapeutic targets, as shown by its successful application in uncovering the role of DRD2 in psoriasis. Beyond psoriasis, this system offers significant potential for exploring pathological mechanisms and discovering therapeutic targets across various diseases.