Equivariant Graph Neural Networks for 3D Macromolecular Structure
This work addresses the problem of analyzing 3D macromolecular structures for researchers in structural biology, representing an incremental improvement over existing equivariant methods.
The paper tackles the challenge of representing 3D macromolecular structures by applying equivariant graph neural networks to structural biology tasks, achieving top performance on three out of eight ATOM3D benchmark tasks and competitive results on others.
Representing and reasoning about 3D structures of macromolecules is emerging as a distinct challenge in machine learning. Here, we extend recent work on geometric vector perceptrons and apply equivariant graph neural networks to a wide range of tasks from structural biology. Our method outperforms all reference architectures on three out of eight tasks in the ATOM3D benchmark, is tied for first on two others, and is competitive with equivariant networks using higher-order representations and spherical harmonic convolutions. In addition, we demonstrate that transfer learning can further improve performance on certain downstream tasks. Code is available at https://github.com/drorlab/gvp-pytorch.