DREAMS: Drug REpurposing with Artificial Intelligence for Muscular Disorders

Rare diseases affect over 300 million people worldwide, yet only 6% currently have an approved treatment option. Given these numbers, and the pharmaceutical industry’s traditional “one drug, one disease” paradigm, it is crucial to develop disruptive approaches that accelerate drug discovery while reducing development time and costs. To address this challenge, the DREAMS project aims to identify repurposable treatments for multiple rare diseases by targeting shared pathological phenotypes. Our approach is built on four key pillars: (i) disease modeling using patient-derived pluripotent stem cells to identify common pathological phenotypes; (ii) phenotypic screening to identify compounds that rescue these molecular and cellular dysfunctions; (iii) artificial intelligence-based target deconvolution to understand mechanisms of action, predict therapeutic targets, and propose new disease indications; and (iv) innovative clinical trial methodologies designed to address groups of diseases rather than single conditions. Importantly, DREAMS is deeply patient-centric, with patient representatives actively involved in the design of the clinical trial to ensure that the research aligns with patients’ needs and priorities. To develop this approach, DREAMS specifically focuses on rare neuromuscular disorders that share a key pathological mechanism: impaired autophagy in striated muscle tissue. Five genetic diseases were selected based on their clinical relevance and preclinical evidence demonstrating that autophagy rescue can be beneficial: dynamin 2-related centronuclear myopathy, Duchenne muscular dystrophy, LMNA-related Emery–Dreifuss muscular dystrophy type 2, glycogen storage disease type II (Pompe disease), and Danon disease. By focusing on shared mechanisms rather than disease-specific mutations, DREAMS establishes a scalable and reusable platform for drug discovery that can be applied to additional rare diseases with overlapping pathophysiological features.