FlexRibbon: Joint Sequence and Structure Pretraining for Protein Modeling
Published in ICLR 2026, 2025
Protein foundation models have advanced rapidly, with most approaches falling into two dominant paradigms. We present FlexRibbon, a pretrained protein model that jointly learns from amino acid sequences and three-dimensional structures. Our pretraining strategy combines masked language modeling with diffusion-based denoising, enabling bidirectional sequence-structure learning without requiring MSAs. Evaluated across diverse tasks, FlexRibbon establishes new state-of-the-art performance on 12 different tasks.
Recommended citation: Jianwei Zhu, Yu Shi, ..., Chenao Li, ..., Tao Qin. (2025). "FlexRibbon: Joint Sequence and Structure Pretraining for Protein Modeling." ICLR 2026.