Detection of gene-environment interactions in the presence of linkage disequilibrium and noise by using genetic risk scores with internal weights from elastic net regression

Table 2 Real data application – ER stress x air pollution interaction in the SALIA study

	p-value PM_2.5	p-value PM_2.5 absorbance	p-value PM₁₀	p-value NO₂
Best SNP¹ (raw p-value²)	0.016	0.040	0.064	0.012
Best SNP¹ (Bonferroni-corrected p-value)	0.130	0.316	0.516	0.095
Weighted GRS³ (p-value GRSxE term)	0.014	0.063	0.102	0.078
Unweighted GRS (p-value GRSxE term)	0.038	0.062	0.249	0.122

Interaction between air pollution exposure and eight SNPs of the endoplasmatic reticulum (ER) stress pathway on the levels of leukotriene (LT) B₄ measured in induced sputum (low vs. high, cut point at 3rd quartile) in 402 women from the SALIA study (p-values are given for the GxE interaction). Air pollution exposures: PM_2.5: fine inhalable particles, with diameters that are generally 2.5 μm and smaller; PM_2.5 absorbance: filter absorbance of PM_2.5 (soot); PM₁₀: inhalable particles, with diameters that are generally 10 μm and smaller; NO₂: Nitrogen dioxide
All models were adjusted for age, BMI (kg/m²), smoking history, passive smoking, level of education and indoor air pollution (heating with fossil fuels and exposure to indoor mold)
¹: The “best SNP” (additive model) had the lowest p-value for the marginal genetic effect as well as for the GxE interaction term (rs2254958, compare Hüls et al. [12])
²: P-values derived from individual SNP by exposure interaction estimates, not corrected for the number of SNPs tested
³: weights were estimated by applying a lasso regression on the combined marginal genetic effects of all eight SNPs on the binary health outcome (low vs. high levels of leukotriene (LT) B₄)

ISSN: 2730-6844