PROPS

PROPS computes probabilistic pathway scores from gene expression data to produce pathway-level features for supervised disease classification by modeling gene interactions with Bayesian networks and probabilistic graphical models.

Key Features:

• Probabilistic Pathway Scoring: Aggregates gene expression values into pathway-level scores using probabilistic graphical models.
• Bayesian Network Representation: Represents biological pathways as Bayesian networks to capture gene interactions and dependencies.
• Pathway-Based Classification: Produces pathway-centric features that emphasize biological mechanisms rather than individual genes for classification tasks.
• Robustness Against Noise: Uses aggregated pathway features to reduce sensitivity to noise inherent in individual gene expression measurements.
• Supervised Machine Learning Integration: Integrates individualized pathway scores into a supervised machine learning framework for disease classification.

Scientific Applications:

• Disease Differentiation and Prediction: Applied to distinguish ulcerative colitis (UC) from Crohn's disease (CD) using gene expression data.
• Enhanced Performance in IBD Classification: Outperformed gene-based and alternative pathway-based classifiers across five IBD datasets when classifying CD versus UC.

Methodology:

Represents biological pathways as Bayesian networks (probabilistic graphical models), aggregates gene expression values into individualized pathway scores, and applies supervised machine learning for classification.