pathDESeq

pathDESeq identifies differentially expressed genes by applying a three-state Markov Random Field model that incorporates biological networks to improve detection sensitivity and interpret RNA-seq normalized count data (FPKM/RPKM) and to aid interpretation with Gene Ontology and KEGG pathway context.

Key Features:

• Three-State Markov Random Field (MRF): Employs a three-state MRF model that incorporates known biological networks to detect differentially expressed genes while reducing false discovery rates (FDR).
• Integration with Biological Networks: Utilizes existing biological pathway databases to account for gene interactions and pathway structure during differential expression analysis.
• Input Requirements: Accepts normalized RNA-seq count data formats such as Fragments Per Kilobase of transcript per Million mapped reads (FPKM) and Reads Per Kilobase of transcript per Million mapped reads (RPKM).
• Performance Superiority: Simulation studies demonstrate improved performance over two-state MRF models and higher sensitivity than DESeq, EBSeq, edgeR, and NOISeq at comparable FDR levels.
• Enhanced Biological Insight: Applied to colorectal cancer and hepatocellular carcinoma datasets, it identifies more significant Gene Ontology terms and KEGG pathways.

Scientific Applications:

• Network-context differential expression analysis: Characterizes gene expression changes while accounting for gene interactions and pathway structure.
• Cancer genomics: Identifies pathway alterations in colorectal cancer and hepatocellular carcinoma relevant to disease progression or treatment response.
• Hypothesis generation: Supports generation of hypotheses about gene function and interaction across biological contexts using enriched GO terms and KEGG pathways.

Methodology:

Implements a three-state Markov Random Field model that incorporates known biological networks on normalized RNA-seq counts (FPKM/RPKM), with performance evaluated by simulation studies comparing to two-state MRF, DESeq, EBSeq, edgeR, and NOISeq and by application to real datasets to assess Gene Ontology and KEGG pathway enrichment.