PINCAGE

PINCAGE integrates RNA-seq gene expression and Illumina 450K array DNA methylation across promoters and gene bodies using probabilistic models to identify genes perturbed by combined genomic and epigenomic alterations and to predict cancer outcomes.

Key Features:

• Probabilistic Integration: Employs probabilistic models to jointly analyze RNA-seq gene expression and 450K array DNA methylation, capturing dependencies between datasets rather than assuming independence.
• Modular Graphical Models: Utilizes modular graphical models to represent relationships between expression and methylation across promoters and gene bodies and to enable incorporation of additional data types.
• Enhanced Biomarker Discovery: Demonstrated on Breast Invasive Carcinoma datasets from The Cancer Genome Atlas (TCGA) consortium to identify candidate development biomarkers PTF1A, RABIF, RAG1AP1, TIMM17A, LOC148145 and progression biomarkers SERPINE3, ZNF706.
• Improved Discrimination: Provides better discrimination between normal and tumor tissues and between progressing and non‑progressing tumors compared with methods that assume data-type independence, especially when integrating evidence across multiple genes.

Scientific Applications:

• Cancer Genomics: Integrated analysis of genomic and epigenomic measurements to uncover genes perturbed in cancer and to inform disease-state prediction.
• Biomarker Identification: Discovery of molecular biomarkers for cancer development and progression, exemplified by candidates identified in TCGA Breast Invasive Carcinoma datasets.
• General Genomic Diseases: Application to other genomic diseases where matched RNA-seq and DNA methylation (450K array) data across promoters and gene bodies are available.

Methodology:

PINCAGE integrates RNA-seq and 450K array DNA methylation using a probabilistic framework implemented with modular graphical models that account for interdependencies between data types.