Genetic risk score (GRS) analysis and classification
In order to compute the polygenic effects of multiple SNVs, genetic risk scores (GRS) were calculated using summary statistics from the largest GWAS of childhood asthma published to date.(15) Our GRS used the pruning and thresholding approach as previously described (32): 1) Identified shared SNVs between the discovery (published GWAS summary statistics) and target (CHILD study) cohorts; 2) Eliminated redundant signals by pruning SNVs in high linkage disequilibrium (LD) using a window size of 50kb, shift distance of 2, and LD threshold (R2) of > 0.8; 3) Employed a stepwise, forward regression analysis to calculate GRS for each individual, starting with the most significant SNV from the published GWAS, by summing the risk alleles weighted by the effect size (β) obtained from the discovery cohort; 4) Identified a set of genetic variants that best predicted the target trait using regression analysis while accounting for sex and the first three PCs as covariates.
Subjects were classified into high (1), moderate (0) and low (-1) GRS groups: defined as high if their GRS was 1 standard deviation higher from the mean (z score > 1), and low if 1 standard deviation lower than the mean (z score < -1). Individuals who fall between high and low risk group ( -1 < z score < 1) were defined as having a moderate risk.