PROTAC-Splitter: A machine learning framework for automated identification of PROTAC substructures
Recommended citation: Ribes, Stefano et al. (2025). "PROTAC-Splitter: A machine learning framework for automated identification of PROTAC substructures." ChemRxiv. link
Introduces PROTAC-Splitter for automated annotation of E3 ligase ligand, linker, and warhead. Trains on a ~1.3M synthetic PROTAC dataset and compares a Transformer seq2seq model with a graph-based XGBoost approach; proposes a reliable hybrid method and releases code/data.
Recommended citation: Ribes, Stefano et al. (2025). “PROTAC-Splitter: A machine learning framework for automated identification of PROTAC substructures.” ChemRxiv.