Demystifying Multimodal Biomolecular Co-design With Intrinsic Geodesic Coupling
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866917553461264384 |
|---|---|
| author | Qiu, Keyue Wang, Xintong Zhang, Zhilong Zhou, Hao Ma, Wei-Ying |
| author_facet | Qiu, Keyue Wang, Xintong Zhang, Zhilong Zhou, Hao Ma, Wei-Ying |
| contents | Biomolecules such as proteins and small-molecule ligands play a central role in biological systems, arising from the tight interplay between sequence and three-dimensional structure. Recent generative models for biomolecular co-design aim to capture this interplay by jointly modeling coupled modalities. However, existing approaches largely adopt a parallel execution of marginal generative processes, implicitly enforcing fixed synchronous coupling. We argue that a critical but overlooked degree of freedom lies in how these marginal processes are temporally coupled during training and generation, where inappropriate coupling can introduce high-variance supervision and inconsistent intermediate states, affecting modality consistency. To address this, we introduce GeoCoupling, a systematic framework that optimizes for temporal couplings between heterogeneous modalities. Empirical results across structure-based drug design and unconditional protein design demonstrate the learned couplings consistently outperform synchronous and randomly coupled baselines, yielding biomolecules with improved physical validity and diversity. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2606_01628 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Demystifying Multimodal Biomolecular Co-design With Intrinsic Geodesic Coupling Qiu, Keyue Wang, Xintong Zhang, Zhilong Zhou, Hao Ma, Wei-Ying Biomolecules Artificial Intelligence Biomolecules such as proteins and small-molecule ligands play a central role in biological systems, arising from the tight interplay between sequence and three-dimensional structure. Recent generative models for biomolecular co-design aim to capture this interplay by jointly modeling coupled modalities. However, existing approaches largely adopt a parallel execution of marginal generative processes, implicitly enforcing fixed synchronous coupling. We argue that a critical but overlooked degree of freedom lies in how these marginal processes are temporally coupled during training and generation, where inappropriate coupling can introduce high-variance supervision and inconsistent intermediate states, affecting modality consistency. To address this, we introduce GeoCoupling, a systematic framework that optimizes for temporal couplings between heterogeneous modalities. Empirical results across structure-based drug design and unconditional protein design demonstrate the learned couplings consistently outperform synchronous and randomly coupled baselines, yielding biomolecules with improved physical validity and diversity. |
| title | Demystifying Multimodal Biomolecular Co-design With Intrinsic Geodesic Coupling |
| topic | Biomolecules Artificial Intelligence |
| url | https://arxiv.org/abs/2606.01628 |