Abstract:Molecular property prediction holds significant importance in drug discovery, enabling the identification of biologically active compounds with favorable drug-like properties. However, the low data problem, arising from the scarcity of labeled data in drug discovery, poses a substantial obstacle for accurate predictions. To address this challenge, we introduce a novel architecture, AttFPGNN-MAML, for few-shot molecular property prediction. The proposed approach incorporates a hybrid feature...

What problem does this paper attempt to address?

The paper is primarily dedicated to addressing the problem of molecular property prediction in the field of drug discovery, particularly in situations where data is scarce (low data volume). Specifically, the study proposes a new architecture called AttFPGNN-MAML, which aims to accurately predict molecular properties using a small number of samples (i.e., in a few-shot learning scenario). Key points mentioned in the paper include: 1. **Background and Challenges**: In the drug discovery process, it is crucial to identify bioactive compounds with favorable pharmacokinetic properties (such as absorption, distribution, metabolism, excretion, and toxicity). However, due to high experimental costs and difficulties in data collection, the amount of data available for training is usually very limited, which constitutes a major obstacle to improving the performance of deep learning models. 2. **Proposed Method**: To address the above issues, the authors propose the AttFPGNN-MAML method, which combines Graph Neural Networks (GNN) with hybrid fingerprint representations. This method utilizes a meta-learning strategy called Prototype Meta-Learning (ProtoMAML) to adapt to new tasks and employs an Instance Attention mechanism to obtain task-specific molecular representations. 3. **Evaluation and Results**: The method was evaluated on two few-shot datasets (MoleculeNet and FS-Mol), and the results showed that the proposed method significantly improved prediction performance compared to existing methods in most cases. Particularly, with smaller support set sizes (e.g., 16, 32, 64 samples), the method performed excellently. 4. **Conclusion and Future Work**: The authors summarize the effectiveness of the method and point out future research directions, including improving the interpretability of the model and validation in real-world projects. In summary, the paper proposes an effective few-shot learning solution for the problem of molecular property prediction in the field of drug discovery, aiming to achieve high-accuracy predictions with minimal data.

Meta Learning with Attention Based FP-GNNs for Few-Shot Molecular Property Prediction

Chemical Property Relation Guided Few-Shot Molecular Property Prediction

Graph Sampling-based Meta-Learning for Molecular Property Prediction

LGGA-MPP: Local Geometry-Guided Graph Attention for Molecular Property Prediction

Meta Learning With Graph Attention Networks for Low-Data Drug Discovery

Few-shot learning with transformers via graph embeddings for molecular property prediction

FP-GNN: a versatile deep learning architecture for enhanced molecular property prediction

Property-guided Few-Shot Learning for Molecular Property Prediction with Dual-View Encoder and Relation Graph Learning Network

Few-shot molecular property prediction via Hierarchically Structured Learning on Relation Graphs

Prompting Meta-Learned Hierarchical Graph Network for Molecular Property Prediction

Attribute-guided prototype network for few-shot molecular property prediction

Property-Aware Relation Networks for Few-Shot Molecular Property Prediction

Few-shot learning via graph embeddings with convolutional networks for low-data molecular property prediction

MolecularGPT: Open Large Language Model (LLM) for Few-Shot Molecular Property Prediction

Adaptive Transfer of Graph Neural Networks for Few-Shot Molecular Property Prediction

HiGNN: Hierarchical Informative Graph Neural Networks for Molecular Property Prediction Equipped with Feature-Wise Attention

Meta-Learning GNN Initializations for Low-Resource Molecular Property Prediction

Knowledge-enhanced Relation Graph and Task Sampling for Few-shot Molecular Property Prediction

In-Context Learning for Few-Shot Molecular Property Prediction

Directed Message Passing Based on Attention for Prediction of Molecular Properties