MOL-AE: Auto-Encoder Based Molecular Representation Learning With 3D Cloze Test Objective

Abstract 3D molecular representation learning has gained tremendous interest and achieved promising performance in various downstream tasks. A series of recent approaches follow a prevalent framework: an encoder-only model coupled with a coordinate denoising objective. However, through a series of analytical experiments, we prove that the encoderonly model with coordinate denoising objective exhibits inconsistency between pre-training and downstream objectives, as well as issues with disrupted atomic identifiers. To address these two issues, we propose M ol -AE for molecular representation learning, an auto-encoder model using positional encoding as atomic identifiers. We also propose a new training objective named 3D Cloze Test to make the model learn better atom spatial relationships from real molecular substructures. Empirical results demonstrate that M ol -AE achieves a large margin performance gain compared to the current state-of-the-art 3D molecular modeling approach. The source codes of M ol -AE are publicly available at https://github.com/yjwtheonly/MolAE .