The Mechanistic Universe: A Direction-Aware Map of Disease Proteins and Drug Candidates for Predicting Therapeutic Effects and Risks

This dataset accompanies the manuscript“The Mechanistic Universe: A Direction-Aware Map of Disease Proteins and Drug Candidatesfor Predicting Therapeutic Effects and Risks.” It provides structured, machine-readable mappings between:• Human disease mechanisms derived from OMIM (gain- and loss-of-function)• Drug–protein interactions curated from ChEMBL• Pathway topology from Reactome• UniProt protein identifiers and functional context The dataset includes:1) Corrective (Class-1) drug–protein–pathway interactions2) Mechanistically misaligned (anti-corrective) drug–protein interactions3) Disease-associated proteins supported by OMIM without approved drugs4) Reactome-annotated proteins lacking OMIM disease annotation All mechanistic directionality assignments are derived algorithmicallyusing GOF/LOF logic and do not rely on clinical outcome data.