Prediction of missing common genes for disease pairs using network based module separation on incomplete human...

Area of Science:

• Genomics
• Systems Biology
• Computational Biology

Background:

• Identifying shared genes in comorbid diseases is key to understanding their underlying biological mechanisms.
• Comorbidities present complex challenges in disease pathogenesis and treatment.
• Current methods may not fully capture the intricate genetic links between co-occurring diseases.

Purpose of the Study:

• To develop a novel computational method for predicting missing common genes between pairs of diseases.
• To enhance the understanding of shared genetic factors in comorbid conditions.
• To facilitate the identification of potential therapeutic targets for complex diseases.

Main Methods:

• Formulating the prediction of missing common genes as an optimization problem.
• Minimizing network-based module separation using gene-disease mappings onto the interactome.
• Utilizing graph-based approaches to analyze gene networks.

Main Results:

• The novel method achieved a high average Receiver Operating Characteristic (ROC) score of 0.95.
• This significantly outperforms a baseline method using randomized data, which yielded an ROC score of 0.60.
• Validation was performed across over 600 disease pairs, demonstrating robust performance.

Conclusions:

• The developed method effectively predicts missing common genes, aiding in the completion of gene sets for comorbid diseases.
• This facilitates a deeper understanding of disease biology and supports the development of gene-targeted therapeutics for comorbid conditions.
• Future work can extend this approach to predict missing network connections (edges) for a more comprehensive disease subgraph analysis.