Other formats:
BibTeX
LaTeX
RIS
@inproceedings{2396557, author = {Bushuiev, Anton and Bushuiev, Roman and Kouba, Petr and Filkin, Anatolii and Gabrielova, Marketa and Gabriel, Michal and Sedlar, Jiri and Pluskal4, Tomas and Damborský, Jiří and Mazurenko, Stanislav and Sivic, Josef}, booktitle = {12th International Conference on Learning Representations 2024}, keywords = {protein-protein interactions; protein design; generalization; self-supervised learning; equivariant 3D representations}, howpublished = {elektronická verze "online"}, language = {eng}, title = {Learning to design protein-protein interactions with enhanced generalization}, url = {https://openreview.net/forum?id=xcMmebCT7s}, year = {2024} }
TY - JOUR ID - 2396557 AU - Bushuiev, Anton - Bushuiev, Roman - Kouba, Petr - Filkin, Anatolii - Gabrielova, Marketa - Gabriel, Michal - Sedlar, Jiri - Pluskal4, Tomas - Damborský, Jiří - Mazurenko, Stanislav - Sivic, Josef PY - 2024 TI - Learning to design protein-protein interactions with enhanced generalization KW - protein-protein interactions KW - protein design KW - generalization KW - self-supervised learning KW - equivariant 3D representations UR - https://openreview.net/forum?id=xcMmebCT7s N2 - Discovering mutations enhancing protein-protein interactions (PPIs) is critical for advancing biomedical research and developing improved therapeutics. While machine learning approaches have substantially advanced the field, they often struggle to generalize beyond training data in practical scenarios. The contributions of this work are three-fold. First, we construct PPIRef, the largest and non-redundant dataset of 3D protein-protein interactions, enabling effective large-scale learning. Second, we leverage the PPIRef dataset to pre-train PPIformer, a new SE(3)-equivariant model generalizing across diverse protein-binder variants. We fine-tune PPIformer to predict effects of mutations on protein-protein interactions via a thermodynamically motivated adjustment of the pre-training loss function. Finally, we demonstrate the enhanced generalization of our new PPIformer approach by outperforming other state-of-the-art methods on new, non-leaking splits of standard labeled PPI mutational data and independent case studies optimizing a human antibody against SARS-CoV-2 and increasing the thrombolytic activity of staphylokinase. ER -
BUSHUIEV, Anton, Roman BUSHUIEV, Petr KOUBA, Anatolii FILKIN, Marketa GABRIELOVA, Michal GABRIEL, Jiri SEDLAR, Tomas PLUSKAL4, Jiří DAMBORSKÝ, Stanislav MAZURENKO and Josef SIVIC. Learning to design protein-protein interactions with enhanced generalization. Online. In \textit{12th International Conference on Learning Representations 2024}. 2024, 26 pp.
|