Integrative analysis of genetical genomics data incorporating network structures

Area of Science:

• Genomics
• Systems Biology
• Bioinformatics

Background:

• Gene regulatory networks (GRNs) are crucial for understanding cellular functions.
• Integrating GRN structures can improve gene selection accuracy for phenotypes.
• Gene expression acts as an intermediate phenotype between genetic variants and traits.

Purpose of the Study:

• To develop a method for accurate gene selection using gene expression data and GRN structures.
• To address endogeneity in gene expression data using genetic variants as instrumental variables.
• To enhance the efficiency of gene selection and estimation in complex biological systems.

Main Methods:

• Proposed a two-step instrumental variable regression approach.
• Utilized the LASSO algorithm in the first step to estimate genetic variant effects on gene expression.
• Employed a graph-constrained regularization method in the second step with projected expression measurements.

Main Results:

• The proposed method demonstrates theoretical selection consistency and provides derived bounds for estimates.
• Simulation studies and real data analyses validated the method's effectiveness.
• The approach outperformed existing methods in gene selection and estimation accuracy.

Conclusions:

• Integrating gene regulatory network information significantly improves gene selection accuracy.
• The novel two-step graph-constrained method offers a robust approach for genetic association studies.
• This work provides a valuable tool for dissecting complex genetic architectures and identifying phenotype-associated genes.