A Gene Set-Integrated Approach for Predicting Disease-Associated Genes

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Identifying disease-associated genes is crucial for understanding complex disease pathogenesis.
• Current disease gene prediction models primarily rely on molecular expression data and networks.
• Existing methods often overlook the valuable information within annotated gene sets and pathways.

Purpose of the Study:

• To develop a novel computational approach for predicting disease-associated genes.
• To integrate annotated gene set data from the Molecular Signature Database (MSigDB) into gene prediction models.
• To enhance the accuracy of disease gene prediction by leveraging pathway and functional gene set information.

Main Methods:

• A new method was developed to predict disease-associated genes by integrating annotated gene sets from MSigDB.
• Annotated gene sets were represented and integrated into a signal matrix, serving as gene features.
• The signal matrix was used to train a disease gene prediction model, which was compared against existing methods.

Main Results:

• The proposed method demonstrated superior performance in predicting genes associated with five complex diseases compared to existing approaches.
• A case study on autism spectrum disorder (ASD) revealed significant associations between predicted genes and ASD.
• Statistical analysis of biological networks and independent ASD studies corroborated the prediction findings.

Conclusions:

• Integrating annotated gene sets significantly improves disease gene prediction accuracy.
• The developed method offers a powerful tool for identifying novel disease-associated genes.
• The findings provide valuable insights into the genetic underpinnings of complex diseases like ASD.