Prediction of Drug-Gene Interaction by Using Metapath2vec

Area of Science:

• Biomedical research
• Bioinformatics
• Genomics

Background:

• Heterogeneous information networks (HINs) are crucial in various fields, including biomedical research.
• Predicting drug-gene and drug-target interactions is vital for understanding drug efficacy and human genomics.
• Existing machine learning and statistical models have limitations in exploring these complex biological network interactions.

Purpose of the Study:

• To introduce and evaluate representation learning methods, specifically metapath2vec and metapath2vec++, for predicting drug-gene relationships.
• To integrate adverse drug reaction (ADR) data and causal relationships into the drug-gene network for improved prediction.
• To compare the performance of these novel methods against established prediction algorithms.

Main Methods:

• Application of metapath2vec and metapath2vec++ representation learning models on a biological heterogeneous network.
• Integration of drug-gene network data with adverse drug reaction (ADR) information.
• Utilizing the skip-gram model for natural language processing-inspired representation learning.
• Employing kernelized Bayesian matrix factorization for prediction completion.
• Comparative analysis with Katz, CATAPULT, and matrix factorization methods.

Main Results:

• Metapath2vec and metapath2vec++ models demonstrated predictive capabilities for drug-gene relationships.
• Performance evaluation using receiver operating characteristic (ROC) curves and area under the curve (AUC) values.
• Comparison highlighted the effectiveness of the proposed representation learning approaches.

Conclusions:

• Representation learning, particularly metapath2vec and metapath2vec++, offers a promising approach for predicting drug-gene interactions in complex biological networks.
• The integration of ADR data enhances the predictive power of these models.
• These findings contribute to advancing drug efficacy research and human genomics through improved biological network analysis.