Gate-guided and subgraph-aware Bilateral Fusion for Molecular Property Prediction
Keywords: Molecular property prediction
TL;DR: We propose a novel method to realize molecular properties prediction task.
Abstract: Predicting molecular properties is crucial in scientific research and industry applications. Molecules are often modeled as graphs where atoms and chemical bonds are represented as nodes and edges, respectively, and Graph Neural Networks (GNNs) have been commonly utilized to predict atom-related properties, such as reactivity and solubility. However, some properties, such as efficacy, and metabolic properties, are closely related to functional groups (subgraphs), which cannot be solely determined by individual atoms. In this paper, we introduce the Gate-guided and Subgraph-aware Bilateral Fusion (GSBF) model for molecular property prediction. GSBF overcomes the limitations of prior atom-wise and subgraph-wise models by integrating both types of information into two distinct branches within the model. We provide a gate-guided mechanism to control the utilization of two branches. Considering existing atom-wise GNNs cannot properly extract invariant subgraph features, we propose a decomposition-polymerization GNN architecture for the subgraph-wise branch. Furthermore, we propose cooperative node-level and graph-level self-supervised learning strategies for GSBF to improve its generalization. Our method offers a more comprehensive way to learn representations for molecular property prediction. Extensive experiments have demonstrated the effectiveness of our method.
Primary Area: applications to physical sciences (physics, chemistry, biology, etc.)
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2024/AuthorGuide.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors' identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 494
Loading