New Software Predicts Which Drug Designs Will Actually Work Before Making Them
Researchers have created COMPASS, a computational tool that identifies which molecular linkers will successfully enable PROTACs—a promising new class of drugs that make cancer cells destroy themselves. By filtering out failing designs before costly lab synthesis, the software could slash development timelines and costs for biotech companies racing to bring targeted therapies to market.
Originaltitel: COMPASS: A Computational Pipeline to Identify Linkers Predicting Ubiquitinable PROTAC-Induced Ternary Complexes.
PROTAC-läkemedel designades för att rikta in sig på och bryta ned sjukdomsrelaterade proteiner, men utvecklingen strandas ofta på linkergeometri. COMPASS är en beräkningsmodell som screnar linkersekvenser innan syntes, vilket sparar resurser och försnabbar utvecklingen. Modellen klassificerar vilka linker-kombinationer som kan bilda funktionella ternära komplex mellan E3-ligaser och målproteiner. Forskarna testade COMPASS mot 20 kristallografiska strukturer med submånometer-noggrannhet och validerade retrospektivt mot 112 redan syntetiserade PROTAC-molekyler från åtta E3/POI-system. Modellen uppnådde 93 procent korrekt klassificering av degraderande PROTAC:er. Institutioner i Rennes, Eindhoven, Göteborg och Rennes samarbetade i studien, finansierad bland annat av Vetenskapsrådet och Svenska Cancerfonden. För bolag som utvecklar PROTAC:er minskar detta utrustningens och försöksomkostnaderna väsentligt och förkortar vägen till kandidatval.
PROteolysis TArgeting Chimeras (PROTACs) are bifunctional molecules designed to induce targeted protein degradation by forming a transient ternary complex between an E3 ubiquitin ligase and a protein of interest (POI), leading to the E3-mediated ubiquitination of the POI and its subsequent proteasomal degradation. Although PROTACs have emerged as highly promising therapeutic tools, rational design remains challenging due to limited structural understanding of the resulting assemblies, the dynamic nature of the ternary interface, and the critical role of the linker. Herein, we present COMPASS (COmputational Modeling of PROTAC Assembly with Structure-based Screening), a computational pipeline that allows the screening of linker libraries by assessing both ternary complex formation and ubiquitination potential. COMPASS functions as a high-sensitivity negative filter, identifying linkers that cannot form productive complexes and enabling their elimination before synthesis. Benchmarking against 20 crystallographic structures yielded <6 Å Cα-RMSD across all systems, outperforming existing methods. Retrospective validation across 8 distinct E3/POI systems (112 PROTACs) yielded 93% recall against degradation endpoints. Discriminative power is strongest when linker geometry is rate-limiting, a regime complementary to the stability and cooperativity effects that static structural modeling cannot capture.