Inferring pathway activity toward precise disease classification.

Area of Science:

• Bioinformatics
• Systems Biology
• Computational Biology

Background:

• Microarray technology enables disease classification via gene expression profiles.
• Classifying complex diseases is challenging due to cellular and genetic heterogeneity.
• Incorporating pathway information offers a promising approach to overcome these challenges.

Purpose of the Study:

• To propose a novel classification method based on inferred patient pathway activities.
• To improve disease classification accuracy by analyzing pathway activity rather than individual gene expression.
• To identify condition-responsive genes (CORGs) within pathways for optimal discriminative power.

Main Methods:

• Developed a new classification method utilizing inferred pathway activities for each patient.
• Summarized pathway activity from gene expression levels of condition-responsive genes (CORGs).
• Evaluated classifier performance against individual gene expression and previous pathway-based methods.

Main Results:

• Classifiers based on pathway activity demonstrated superior performance compared to individual gene expression classifiers.
• The method achieved better results in both simple and complex case-control studies, including cell differentiation and cancer subtyping.
• The proposed approach outperformed existing methods that use static pathway definitions.

Conclusions:

• Pathway activity-based classification is effective for complex diseases, outperforming individual gene expression analysis.
• The identified CORGs can serve as potential diagnostic markers and reveal core disease alterations.
• This method offers a robust framework for disease classification and biomarker discovery.