DNA-based eyelid trait prediction in Chinese Han population

The eyelid folding represents one of the most distinguishing features of East Asian faces, involving the absence or presence of the eyelid crease, i.e., single vs. double eyelid. Recently, a genome-wide association study (GWAS) identified two SNPs (rs12570134 and rs1415425) showing genome-wide significant association with the double eyelid phenotype in Japanese. Here we report a confirmatory study in 697 Chinese individuals of exclusively Han origin. Only rs1415425 was statistically significant (P-value = 0.011), and the allele effect was on the same direction with that reported in Japanese. This SNP combined with gender and age explained 10.0% of the total variation in eyelid folding. DNA-based prediction model for the eyelid trait was developed and evaluated using logistic regression. The model showed mild to moderate predictive capacity (AUC = 0.69, sensitivity = 63%, and specificity = 70%). We further selected six additional SNPs by massive parallel sequencing of 19 candidate genes in 24 samples, and one SNP rs2761882 was statistically significant (P-value = 0.027). All predictors including these two SNPs (rs1415425 and rs2761882), gender, and age explained 11.2% of the total variation. The combined prediction model obtained an improved predictive capacity (AUC = 0.72, sensitivity = 62%, and specificity = 66%). Our study thus provided a confirmation of previous GWAS findings and a DNA-based prediction of the eyelid trait in Chinese Han individuals. This model may add value to forensic DNA phenotyping applications considering gender and age can be separately inferred from genetic and epigenetic markers. To further improve the prediction accuracy, future studies should focus on identifying more informative SNPs by large GWASs in East Asian populations.