Multi-omic Causal Discovery using Genotypes and Gene Expression

Causal inference in genomics is inherently challenging due to high dimensionality, hidden confounders, and the intricate task of distinguishing direct regulatory interactions from indirect ones. We introduce GENESIS (Gene Network Inference from Expression Signals and SNPs), a novel algorithm that fuses genotype (SNP) data with transcriptomic profiles to reconstruct directed gene regulatory networks. Our algorithm adopts a robust two-stage hypothesis testing strategy: (1)it conducts marginal testing of SNP-gene associations within defined genomic windows; (2)it applies conditional independence tests to eliminate indirect regulatory effects. We run our method on a real-world dataset and compare the results. This parametrically rigorous framework offers a powerful avenue for uncovering causal pathways in complex traits, with promising applications in functional genomics, therapeutic interventions, and precision medicine.