Using deep learning to associate human genes with age-related diseases

Area of Science:

• Genomics
• Computational Biology
• Machine Learning

Background:

• Identifying genes associated with aging and age-related diseases is crucial for understanding disease mechanisms.
• Existing methods may not fully leverage complex biological data for gene-disease association prediction.

Purpose of the Study:

• To develop and evaluate a novel Deep Neural Network (DNN) method for classifying genes associated with age-related diseases.
• To identify novel gene-disease associations using machine learning by integrating multiple biological data sources.

Main Methods:

• A novel modular Deep Neural Network (DNN) architecture was designed to integrate various biological descriptors (Gene Ontology terms, protein-protein interactions, pathway information).
• The DNN model was trained on a dataset of human genes associated with age-related diseases.
• Two strategies were employed to identify potential novel gene-disease associations: proximity in the DNN's feature space and high class label probabilities.

Main Results:

• The proposed DNN method demonstrated superior predictive performance compared to standard DNNs, Gradient Boosted Trees, and Logistic Regression classifiers.
• The method successfully identified potential novel associations between genes and age-related diseases, supported by literature evidence.
• The modular DNN architecture effectively combined multiple biological data sources for improved prediction accuracy.

Conclusions:

• The novel DNN approach offers a powerful tool for identifying genes associated with aging and age-related diseases.
• This method enhances the discovery of novel gene-disease links by leveraging integrated biological data.
• The findings contribute to a better understanding of the genetic underpinnings of aging and associated pathologies.