Improved multimodal protein language model-driven universal biomolecules-binding protein design with EiRA

The interactions between proteins and other biomolecules, such as nucleic acids, form a complex system that supports life activities. Designing proteins capable of targeted biomolecular binding is therefore critical for protein engineering and gene therapy. In this study, we propose a new generative model, EiRA, specifically designed for universal biomolecular binding protein design, which undergo two-stage post-training, i.e., domain-adaptive masking training and binding site-informed preference optimization, based on a general multimodal protein language model. A multidimensional evaluation reveals the SOTA performance of EiRA, including structural confidence, diversity, novelty, and designability on eight test sets across six biomolecule types. Meanwhile, EiRA provides a better characterization of biomolecular binding proteins than generic models, thereby improving the predictive performance of various downstream tasks. We also mitigate severe repetition generation in the original language model by optimizing training strategies and loss. Additionally, we introduced DNA information into EiRA to support DNA-conditioned binder design, further expanding the boundaries of the design paradigm.