PRS Construction
We separated the 70% sample size of our cohort to construct the training GWAS summary statistics. PRS-CS24 was used to infer the posterior effect size of SNPs using the IgE summary statistics of the training group. The 1000 genome EAS phase3 samples were used as LD reference panel. We further adopted the effect size in the rest of 30% samples and used PLINK2.025 to sum up the effect allele dosages of each individual and generate the IgE PRS. We checked the correlation between IgE PRS and IgE using Pearson correlation coefficient test. In order to investigate the association between different quantiles of IgE PRS groups and IgE, we divided the IgE PRS into 11 parts according to the IgE PRS quantiles (0-10th, 10-20th, 20-30th, 30-40th, 40-50th, 50-55th, 55-60th, 60-70th, 70-8th, 80-90th, and 90-100th). The population within the 50th to 55th quantile range was selected as the reference group. The logistic regression was performed compared with each group with the reference group and adjust with age, sex, and PC1 to PC10.