DrugTargetWorld provides synthetic biobanks for AI scientist training

DrugTargetWorld is a framework that procedurally generates simulated biobanks with 54,000 participants to train AI agents in drug target discovery. It includes genotypes, proteins, and synthetic MRI data with concealed causal structures to evaluate an agent's ability to identify disease drivers.