Few-shot Molecular Property Prediction via Hierarchically Structured Learning on Relation Graphs

Ju, Wei; Liu, Zequn; Qin, Yifang; Feng, Bin; Wang, Chen; Guo, Zhihui; Luo, Xiao; Zhang, Ming

doi:10.1016/j.neunet.2023.03.034

Cited by 23 publications

(8 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We conduct a comparative analysis of our approach against two categories of baselines across the four few-shot MoleculeNet benchmark tasks. Methods without a pretrained molecular graph encoder, such as HSL-RG – , ADKF-IFT, PAR, IterRefLSTM, EGNN, TPN, MAML, ProtoNet, and Siamese Methods with a pretrained molecular graph encoder, such as Pre-PAR + +MTA, HSL-RG, Pre-GNN, Meta-MGNN, Pre-PAR, and Pre-ADKF-IFT . It is worth noting that all methods within this category employ a pretrained GIN, and the pretrained GIN weights are provided by Hu et al …”

Section: Resultsmentioning

confidence: 99%

“…(1) Methods without a pretrained molecular graph encoder, such as HSL-RG − , 22 ADKF-IFT, 20 PAR, 19 Iter-RefLSTM, 18 EGNN, 38 TPN, 39 MAML, 15 ProtoNet, 14 and Siamese. 40 (2) Methods with a pretrained molecular graph encoder, such as Pre-PAR + +MTA, 23 HSL-RG, 22 Pre-GNN, 41 Meta-MGNN, 17 Pre-PAR, 19 and Pre-ADKF-IFT. 20 It is worth noting that all methods within this category employ a pretrained GIN, and the pretrained GIN weights are provided by Hu et al 41 Evaluation Procedure and Performance.…”

Section: ■ Results and Discussionmentioning

confidence: 99%

“…We conduct a comparative analysis of our approach against two categories of baselines across the four few-shot MoleculeNet benchmark tasks. Methods without a pretrained molecular graph encoder, such as HSL-RG – , 22 ADKF-IFT, 20 PAR, 19 IterRefLSTM, 18 EGNN, 38 TPN, 39 MAML, 15 ProtoNet, 14 and Siamese. 40 Methods with a pretrained molecular graph encoder, such as Pre-PAR + +MTA, 23 HSL-RG, 22 Pre-GNN, 41 Meta-MGNN, 17 Pre-PAR, 19 and Pre-ADKF-IFT.…”

Section: Resultsmentioning

confidence: 99%

“…Schimunek et al 21 introduced the MHNfs method, which leverages large context molecules to enrich molecular representations and achieve superior few-shot molecular property prediction performance. More recently, Ju et al 22 proposed hierarchically structured learning on relation graphs (HSL-RG), which constructs global relation graphs and utilizes self-supervised learning to acquire transformation-invariant representations of molecules. Meng et al 23 introduced motif-based task augmentation (MTA) technique to enhance the generalization capability of MAML-based methods, which generate new labeled samples through the retrieval of highly relevant motifs.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

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

Qian,

Ju,

Shen

et al. 2024

ACS Omega

View full text Add to dashboard Cite

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 representation to enrich molecular representations and model intermolecular relationships specific to the task. By leveraging ProtoMAML, a meta-learning strategy, our model is trained and adapted to new tasks. Evaluation on two few-shot data sets, MoleculeNet and FS-Mol, demonstrates our method’s superior performance in three out of four tasks and across various support set sizes. These results convincingly validate the effectiveness of our method in the realm of few-shot molecular property prediction. The source code is publicly available at .

show abstract

Section: Resultsmentioning

confidence: 99%

Section: ■ Results and Discussionmentioning

confidence: 99%

“…We conduct a comparative analysis of our approach against two categories of baselines across the four few-shot MoleculeNet benchmark tasks. Methods without a pretrained molecular graph encoder, such as HSL-RG – , 22 ADKF-IFT, 20 PAR, 19 IterRefLSTM, 18 EGNN, 38 TPN, 39 MAML, 15 ProtoNet, 14 and Siamese. 40 Methods with a pretrained molecular graph encoder, such as Pre-PAR + +MTA, 23 HSL-RG, 22 Pre-GNN, 41 Meta-MGNN, 17 Pre-PAR, 19 and Pre-ADKF-IFT.…”

Section: Resultsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

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

Qian,

Ju,

Shen

et al. 2024

ACS Omega

View full text Add to dashboard Cite

show abstract

“…Unlike sequence-based neural networks or array-based neural networks, these models use not only features but also connection information between nodes to enhance accuracy by extracting more information [ 20 , 21 ].…”

Section: Introductionmentioning

confidence: 99%

GraphMHC: Neoantigen prediction model applying the graph neural network to molecular structure

Jeong,

Cho,

Gim

et al. 2024

PLoS ONE

View full text Add to dashboard Cite

Neoantigens are tumor-derived peptides and are biomarkers that can predict prognosis related to immune checkpoint inhibition by estimating their binding to major histocompatibility complex (MHC) proteins. Although deep neural networks have been primarily used for these prediction models, it is difficult to interpret the models reported thus far as accurately representing the interactions between biomolecules. In this study, we propose the GraphMHC model, which utilizes a graph neural network model applied to molecular structure to simulate the binding between MHC proteins and peptide sequences. Amino acid sequences sourced from the immune epitope database (IEDB) undergo conversion into molecular structures. Subsequently, atomic intrinsic informations and inter-atomic connections are extracted and structured as a graph representation. Stacked graph attention and convolution layers comprise the GraphMHC network which classifies bindings. The prediction results from the test set using the GraphMHC model showed a high performance with an area under the receiver operating characteristic curve of 92.2% (91.9-92.5%), surpassing a baseline model. Moreover, by applying the GraphMHC model to melanoma patient data from The Cancer Genome Atlas project, we found a borderline difference (0.061) in overall survival and a significant difference in stromal score between the high and low neoantigen load groups. This distinction was not present in the baseline model. This study presents the first feature-intrinsic method based on biochemical molecular structure for modeling the binding between MHC protein sequences and neoantigen candidate peptide sequences. This model can provide highly accurate responsibility information that can predict the prognosis of immune checkpoint inhibitors to cancer patients who want to apply it.

show abstract

Task‐Similarity is a Crucial Factor for Few‐Shot Meta‐Learning of Structure‐Activity Relationships

Kötter,

Allenspach,

Grebner

et al. 2024

ChemBioChem

View full text Add to dashboard Cite

Machine learning models support computer‐aided molecular design and compound optimization. However, the initial phases of drug discovery often face a scarcity of training data for these models. Meta‐learning has emerged as a potentially promising strategy, harnessing the wealth of structure‐activity data available for known targets to facilitate efficient few‐shot model training for the specific target of interest. In this study, we assessed the effectiveness of two different meta‐learning methods, namely model‐agnostic meta‐learning (MAML) and adaptive deep kernel fitting (ADKF), specifically in the regression setting. We investigated how factors such as dataset size and the similarity of training tasks impact predictability. The results indicate that ADKF significantly outperformed both MAML and a single‐task baseline model on the inhibition data. However, the performance of ADKF varied across different test tasks. Our findings suggest that considerable enhancements in performance can be anticipated primarily when the task of interest is similar to the tasks incorporated in the meta‐learning process.

show abstract

Few-shot Molecular Property Prediction via Hierarchically Structured Learning on Relation Graphs

Cited by 23 publications

References 38 publications

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

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

GraphMHC: Neoantigen prediction model applying the graph neural network to molecular structure

Task‐Similarity is a Crucial Factor for Few‐Shot Meta‐Learning of Structure‐Activity Relationships

Contact Info

Product

Resources

About