MedGraphNet: Leveraging Multi-Relational Graph Neural Networks and Text Knowledge for Biomedical Predictions

Oladimeji S Macaulay, Michael Servilla, Kushal Virupakshappa, David Arredondo, Yue Hu, Luis Tafoya, Yanfu Zhang, Avinash Sahu
Proceedings of the 19th Machine Learning in Computational Biology meeting, PMLR 261:162-182, 2024.

Abstract

Genetic, molecular, and environmental factors influence diseases through complex interactions with genes, phenotypes, and drugs. Current methods often fail to integrate diverse multi-relational biological data meaningfully, limiting the discovery of novel risk genes and drugs. To address this, we present MedGraphNet, a multi-relational Graph Neural Network (GNN) model designed to infer relationships among drugs, genes, diseases, and phenotypes. MedGraphNet initializes nodes using informative embeddings from existing text knowledge, allowing for robust integration of various data types and improved generalizability. Our results demonstrate that MedGraphNet matches and often outperforms traditional single-relation approaches, particularly in scenarios with isolated or sparsely connected nodes. The model shows generalizability to external datasets, achieving high accuracy in identifying disease-gene associations and drug-phenotype relationships. Notably, MedGraphNet accurately inferred drug side effects without direct training on such data. Using Alzheimer’s disease as a case study, MedGraphNet successfully identified relevant phenotypes, genes, and drugs, corroborated by existing literature. These findings demonstrate the potential of integrating multi-relational data with text knowledge to enhance biomedical predictions and drug repurposing for diseases.

Cite this Paper

BibTeX
@InProceedings{pmlr-v261-macaulay24a, title = {MedGraphNet: Leveraging Multi-Relational Graph Neural Networks and Text Knowledge for Biomedical Predictions}, author = {Macaulay, Oladimeji S and Servilla, Michael and Virupakshappa, Kushal and Arredondo, David and Hu, Yue and Tafoya, Luis and Zhang, Yanfu and Sahu, Avinash}, booktitle = {Proceedings of the 19th Machine Learning in Computational Biology meeting}, pages = {162--182}, year = {2024}, editor = {Knowles, David A and Mostafavi, Sara}, volume = {261}, series = {Proceedings of Machine Learning Research}, month = {05--06 Sep}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v261/main/assets/macaulay24a/macaulay24a.pdf}, url = {https://proceedings.mlr.press/v261/macaulay24a.html}, abstract = {Genetic, molecular, and environmental factors influence diseases through complex interactions with genes, phenotypes, and drugs. Current methods often fail to integrate diverse multi-relational biological data meaningfully, limiting the discovery of novel risk genes and drugs. To address this, we present MedGraphNet, a multi-relational Graph Neural Network (GNN) model designed to infer relationships among drugs, genes, diseases, and phenotypes. MedGraphNet initializes nodes using informative embeddings from existing text knowledge, allowing for robust integration of various data types and improved generalizability. Our results demonstrate that MedGraphNet matches and often outperforms traditional single-relation approaches, particularly in scenarios with isolated or sparsely connected nodes. The model shows generalizability to external datasets, achieving high accuracy in identifying disease-gene associations and drug-phenotype relationships. Notably, MedGraphNet accurately inferred drug side effects without direct training on such data. Using Alzheimer’s disease as a case study, MedGraphNet successfully identified relevant phenotypes, genes, and drugs, corroborated by existing literature. These findings demonstrate the potential of integrating multi-relational data with text knowledge to enhance biomedical predictions and drug repurposing for diseases.} }
Endnote
%0 Conference Paper %T MedGraphNet: Leveraging Multi-Relational Graph Neural Networks and Text Knowledge for Biomedical Predictions %A Oladimeji S Macaulay %A Michael Servilla %A Kushal Virupakshappa %A David Arredondo %A Yue Hu %A Luis Tafoya %A Yanfu Zhang %A Avinash Sahu %B Proceedings of the 19th Machine Learning in Computational Biology meeting %C Proceedings of Machine Learning Research %D 2024 %E David A Knowles %E Sara Mostafavi %F pmlr-v261-macaulay24a %I PMLR %P 162--182 %U https://proceedings.mlr.press/v261/macaulay24a.html %V 261 %X Genetic, molecular, and environmental factors influence diseases through complex interactions with genes, phenotypes, and drugs. Current methods often fail to integrate diverse multi-relational biological data meaningfully, limiting the discovery of novel risk genes and drugs. To address this, we present MedGraphNet, a multi-relational Graph Neural Network (GNN) model designed to infer relationships among drugs, genes, diseases, and phenotypes. MedGraphNet initializes nodes using informative embeddings from existing text knowledge, allowing for robust integration of various data types and improved generalizability. Our results demonstrate that MedGraphNet matches and often outperforms traditional single-relation approaches, particularly in scenarios with isolated or sparsely connected nodes. The model shows generalizability to external datasets, achieving high accuracy in identifying disease-gene associations and drug-phenotype relationships. Notably, MedGraphNet accurately inferred drug side effects without direct training on such data. Using Alzheimer’s disease as a case study, MedGraphNet successfully identified relevant phenotypes, genes, and drugs, corroborated by existing literature. These findings demonstrate the potential of integrating multi-relational data with text knowledge to enhance biomedical predictions and drug repurposing for diseases.
APA
Macaulay, O.S., Servilla, M., Virupakshappa, K., Arredondo, D., Hu, Y., Tafoya, L., Zhang, Y. & Sahu, A.. (2024). MedGraphNet: Leveraging Multi-Relational Graph Neural Networks and Text Knowledge for Biomedical Predictions. Proceedings of the 19th Machine Learning in Computational Biology meeting, in Proceedings of Machine Learning Research 261:162-182 Available from https://proceedings.mlr.press/v261/macaulay24a.html.

Related Material