E(3)-equivariant models cannot learn chirality: Field-based molecular generation
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2024
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866910913692434432 |
|---|---|
| author | Dumitrescu, Alexandru Korpela, Dani Heinonen, Markus Verma, Yogesh Iakovlev, Valerii Garg, Vikas Lähdesmäki, Harri |
| author_facet | Dumitrescu, Alexandru Korpela, Dani Heinonen, Markus Verma, Yogesh Iakovlev, Valerii Garg, Vikas Lähdesmäki, Harri |
| contents | Obtaining the desired effect of drugs is highly dependent on their molecular geometries. Thus, the current prevailing paradigm focuses on 3D point-cloud atom representations, utilizing graph neural network (GNN) parametrizations, with rotational symmetries baked in via E(3) invariant layers. We prove that such models must necessarily disregard chirality, a geometric property of the molecules that cannot be superimposed on their mirror image by rotation and translation. Chirality plays a key role in determining drug safety and potency. To address this glaring issue, we introduce a novel field-based representation, proposing reference rotations that replace rotational symmetry constraints. The proposed model captures all molecular geometries including chirality, while still achieving highly competitive performance with E(3)-based methods across standard benchmarking metrics. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2402_15864 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | E(3)-equivariant models cannot learn chirality: Field-based molecular generation Dumitrescu, Alexandru Korpela, Dani Heinonen, Markus Verma, Yogesh Iakovlev, Valerii Garg, Vikas Lähdesmäki, Harri Machine Learning Chemical Physics Biomolecules Obtaining the desired effect of drugs is highly dependent on their molecular geometries. Thus, the current prevailing paradigm focuses on 3D point-cloud atom representations, utilizing graph neural network (GNN) parametrizations, with rotational symmetries baked in via E(3) invariant layers. We prove that such models must necessarily disregard chirality, a geometric property of the molecules that cannot be superimposed on their mirror image by rotation and translation. Chirality plays a key role in determining drug safety and potency. To address this glaring issue, we introduce a novel field-based representation, proposing reference rotations that replace rotational symmetry constraints. The proposed model captures all molecular geometries including chirality, while still achieving highly competitive performance with E(3)-based methods across standard benchmarking metrics. |
| title | E(3)-equivariant models cannot learn chirality: Field-based molecular generation |
| topic | Machine Learning Chemical Physics Biomolecules |
| url | https://arxiv.org/abs/2402.15864 |