pareg

pareg performs pathway enrichment analysis using a regularized generalized linear model that regresses differential expression p-values against a pathway membership matrix while accounting for inter-pathway dependencies arising from shared genes.

Key Features:

• Regularized Generalized Linear Model: PAREG employs a regularized generalized linear model that incorporates inter-pathway dependencies arising from shared genes to improve pathway enrichment analysis.
• Robustness to Noise: The method is engineered to be resilient to noise, producing reliable results in complex datasets.
• Integration of Biological Knowledge: The regularized regression framework integrates additional biological insights into enrichment analysis for more nuanced interpretation of gene expression data.
• Pathway Redundancy Management: PAREG models dependencies between pathways to address redundancy in pathway databases such as KEGG, Reactome, and Gene Ontology.

Scientific Applications:

• Differential gene expression analysis: Maps differential expression p-values to pathway-level enrichment to interpret multi-condition experiments.
• Cancer research: Recovers known biological pathways and aids identification of novel treatment targets in cancer studies.
• TCGA breast cancer analysis: Demonstrated application to breast cancer samples from The Cancer Genome Atlas (TCGA) for pathway recovery and target identification.
• Confirmatory and exploratory studies: Applicable to both confirmatory and exploratory analyses across biomedical research fields.

Methodology:

PAREG regresses differential expression p-values from multi-condition experiments against a pathway membership matrix using a regularized generalized linear model and incorporates term-term relations to model inter-pathway dependencies, enhancing robustness and interpretability of pathway enrichment scores.