Relating drug response to epigenetic and genetic markers using a region-based kernel score test

Area of Science:

• Genetics and Genomics
• Epigenetics
• Biostatistics

Background:

• Triglyceride (TG) levels are influenced by genetic and epigenetic factors.
• Lipid-lowering medication aims to manage TG levels, but individual responses vary.
• Understanding the interplay between genetic and epigenetic markers is crucial for personalized medicine.

Purpose of the Study:

• To investigate the association of genetic regions of interest (ROIs) with log-transformed triglyceride (TG) levels post-lipid-lowering medication.
• To incorporate kernels for cytosine-phosphate-guanine (CpG) markers and compare them with a parametric model.
• To explore the interaction between genetic and epigenetic data for enhanced information gain.

Main Methods:

• Utilized both real (n=150) and simulated (n=111) datasets for analysis.
• Employed kernels for CpG markers and linear regression to assess associations with post-treatment TG levels, adjusting for pre-treatment levels and age.
• Investigated specific gene introns (e.g., CPT1A) and simulated regions containing causal and non-causal markers.

Main Results:

• Single-CpG-marker results using kernels and linear regression showed good agreement in both datasets.
• Hints of association were found for specific CpG sites (cg17058475, cg00574958) with post-treatment TG levels in real data.
• The simulation demonstrated plausible results for CpG kernels across different window sizes and identified effects for SNPs with high heritabilities when interacting with nearby CpG markers.

Conclusions:

• Kernels for CpG markers are a feasible approach for analyzing epigenetic data.
• The integration of genetic and epigenetic data, particularly through interaction models, can enhance information, even in small sample sizes.
• Further development of kernels for joint multi-omics analysis is warranted to reduce multiple testing burdens and improve biological insights.