Transcriptome-Wide Root Causal Inference.

Area of Science:

• Genetics and Genomics
• Systems Biology
• Computational Biology

Background:

• Root causal genes initiate disease processes through early perturbations.
• Identifying these genes is crucial for developing effective disease interventions.
• Current methods cannot discover root causal genes using only observational data.

Purpose of the Study:

• To propose a novel algorithm, Transcriptome-Wide Root Causal Inference (TWRCI), for identifying root causal genes.
• To leverage genetic variants and bulk RNA sequencing data for causal inference.
• To uncover the underlying causal graph and estimate root causal effects.

Main Methods:

• Developed the TWRCI algorithm integrating genetic variant and gene expression data.
• Employed a competitive regression procedure to link genetic variants to directly caused gene expression.
• Simultaneously inferred causal ordering and estimated root causal effects.

Main Results:

• TWRCI successfully identifies root causal genes and their causal graph.
• The algorithm outperforms existing methods by directly targeting root causal genes.
• Demonstrated TWRCI's efficacy in uncovering root causal mechanisms for two complex diseases.

Conclusions:

• TWRCI provides a powerful new approach for discovering root causal genes from observational data.
• This method has significant implications for understanding disease pathogenesis and developing targeted therapies.
• Findings were validated through replication with independent genome-wide summary statistics.