Salvato in:
| Autori principali: | , , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | https://arxiv.org/abs/2507.11818 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866912865077690368 |
|---|---|
| author | Rekesh, Andrei Cretu, Miruna Shevchuk, Dmytro Somnath, Vignesh Ram Liò, Pietro Batey, Robert A. Tyers, Mike Koziarski, Michał Liu, Cheng-Hao |
| author_facet | Rekesh, Andrei Cretu, Miruna Shevchuk, Dmytro Somnath, Vignesh Ram Liò, Pietro Batey, Robert A. Tyers, Mike Koziarski, Michał Liu, Cheng-Hao |
| contents | Synthesizability remains a critical bottleneck in generative molecular design. While recent advances have addressed synthesizability in 2D graphs, extending these constraints to 3D for geometry-based conditional generation remains largely unexplored. In this work, we present SynCoGen (Synthesizable Co-Generation), a single framework that combines simultaneous masked graph diffusion and flow matching for synthesizable 3D molecule generation. SynCoGen samples from the joint distribution of molecular building blocks, chemical reactions, and atomic coordinates. To train the model, we curated SynSpace, a dataset family containing over 1.2M synthesis-aware building block graphs and 7.5M conformers. SynCoGen achieves state-of-the-art performance in unconditional small molecule graph and conformer co-generation. For protein ligand generation in drug discovery, the amortized model delivers superior performance in both molecular linker design and pharmacophore-conditioned generation across diverse targets without relying on any scoring functions. Overall, this multimodal non-autoregressive formulation represents a foundation for a range of molecular design applications, including analog expansion, lead optimization, and direct de novo design. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2507_11818 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | SynCoGen: Synthesizable 3D Molecule Generation via Joint Reaction and Coordinate Modeling Rekesh, Andrei Cretu, Miruna Shevchuk, Dmytro Somnath, Vignesh Ram Liò, Pietro Batey, Robert A. Tyers, Mike Koziarski, Michał Liu, Cheng-Hao Machine Learning Synthesizability remains a critical bottleneck in generative molecular design. While recent advances have addressed synthesizability in 2D graphs, extending these constraints to 3D for geometry-based conditional generation remains largely unexplored. In this work, we present SynCoGen (Synthesizable Co-Generation), a single framework that combines simultaneous masked graph diffusion and flow matching for synthesizable 3D molecule generation. SynCoGen samples from the joint distribution of molecular building blocks, chemical reactions, and atomic coordinates. To train the model, we curated SynSpace, a dataset family containing over 1.2M synthesis-aware building block graphs and 7.5M conformers. SynCoGen achieves state-of-the-art performance in unconditional small molecule graph and conformer co-generation. For protein ligand generation in drug discovery, the amortized model delivers superior performance in both molecular linker design and pharmacophore-conditioned generation across diverse targets without relying on any scoring functions. Overall, this multimodal non-autoregressive formulation represents a foundation for a range of molecular design applications, including analog expansion, lead optimization, and direct de novo design. |
| title | SynCoGen: Synthesizable 3D Molecule Generation via Joint Reaction and Coordinate Modeling |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2507.11818 |