A module-based approach for post-omics, post-GWAS network-based gene classification

Area of Science:

• Genomics and Bioinformatics
• Computational Biology
• Systems Biology

Background:

• Complex traits and diseases are polygenic, posing a challenge for identifying all involved genes.
• Experimental methods like transcriptomics and Genome-Wide Association Studies (GWAS) yield noisy gene lists, potentially missing true disease genes.
• Network-based computational approaches can expand gene lists by leveraging functional gene relationships.

Purpose of the Study:

• To develop a novel network-based gene classification method, ModGenePlexus, to improve disease gene discovery from noisy omics and GWAS data.
• To address the limitations of previous methods like GenePlexus when applied to complex disease gene lists.
• To enhance the interpretability of gene classification results for biological insights.

Main Methods:

• ModGenePlexus employs a two-stage approach: first, clustering and semi-supervised learning to denoise gene lists into network modules.
• Second, it trains supervised classifiers for each module and aggregates predictions for genome-wide rankings.
• The method was benchmarked using simulated data, transcriptomic signatures, and GWAS datasets across numerous diseases.

Main Results:

• ModGenePlexus demonstrated improved recovery of known disease genes compared to GenePlexus across various datasets.
• The method effectively decomposes noisy gene lists into coherent, biologically relevant network modules.
• Enrichment analysis of ModGenePlexus outputs provided more interpretable insights into nuanced biological processes.

Conclusions:

• ModGenePlexus is a scalable and interpretable tool for classifying genes from GWAS and omics data.
• The method enhances the identification of disease-associated genes, overcoming limitations of previous approaches.
• ModGenePlexus facilitates a deeper understanding of the complex genetic architecture of diseases.