Identifying differentially expressed genes in cancer patients using a non-parameter Ising model

Area of Science:

• Systems biology
• Bioinformatics
• Computational biology

Background:

• Identifying genes and pathways in diseases is crucial for systems biology.
• Integrating protein-protein interaction networks and gene expression data is a key challenge.

Purpose of the Study:

• To develop a novel non-parameter Ising model for identifying differentially expressed (DE) genes.
• To integrate protein-protein interaction networks and microarray data for enhanced gene discovery.
• To apply the model to breast cancer data for improved identification of cancer-related genes and sub-networks.

Main Methods:

• Developed a novel non-parameter Ising model.
• Employed a simulated annealing algorithm to optimize the Ising model.
• Applied the model to two breast cancer microarray datasets.
• Compared results with the Markov random field model.

Main Results:

• The Ising model identified more cancer-related differentially expressed (DE) sub-networks and genes than the Markov random field model.
• Cross-validation demonstrated that DE genes identified by the Ising model improved classification performance.
• The model effectively integrates network and expression data for disease gene identification.

Conclusions:

• The novel Ising model is a powerful tool for identifying DE genes and biological pathways in diseases.
• This approach enhances the understanding of disease mechanisms by integrating diverse biological data.
• The method shows promise for improving diagnostic and prognostic tools in cancer research.