Causal Inference Enrichment (CIE)

Causal Inference Enrichment (CIE) infers active transcriptional regulators from differential gene expression by performing directional gene set enrichment within biological regulatory networks.

Key Features:

• Regularized Gaussian Graphical Model (GGM): Constructs a transcriptional regulatory network using a regularized Gaussian Graphical Model to identify TF-gene interactions.
• ChIP-seq and tissue-specific RNA-seq integration: Combines publicly available ChIP-seq data with tissue-specific RNA-seq gene expression profiles to infer high-confidence interactions.
• Mode-of-regulation annotation: Annotates TF-gene interactions to indicate activation or repression.
• Directional gene set enrichment analysis: Performs directional enrichment analysis on differential gene expression to infer regulator activity.
• Database-supported networks: Utilizes networks derived from STRING-DB, TRRUST, and TRED.
• Custom network support via R-package: Accepts custom network inputs through an R-package interface.
• Benchmarking against curated databases: Validates inferred interactions and modes of regulation against manually curated TF-gene interaction databases.

Scientific Applications:

• In vitro overexpression studies: Infers active transcriptional regulators driving measured differential expression in controlled overexpression experiments.
• Stem-cell differentiation: Identifies regulators associated with transcriptional changes during stem-cell differentiation experiments.
• Fibroblast phenotypic plasticity: Investigates transcriptional mechanisms underlying fibroblast phenotypic plasticity.
• Regulatory mechanism elucidation: Elucidates regulatory mechanisms underlying specific molecular and environmental perturbations.
• Method benchmarking: Assesses accuracy of inferred modes of regulation by comparison to curated TF-gene interaction databases.

Methodology:

Constructs transcriptional regulatory networks using a regularized Gaussian Graphical Model (GGM) that integrates ChIP-seq and tissue-specific RNA-seq, annotates TF–gene interactions as activation or repression, performs directional gene set enrichment analysis, uses networks from STRING-DB, TRRUST and TRED, accepts custom networks via an R-package, and benchmarks results against curated TF-gene databases.