Determination of human facial morphology based on snp markers associated with eye and forehead structures

In cases of unsolved crimes or mass disasters, the genetic prediction of an individual's physical appearance can substantially aid in narrowing the pool of suspects and facilitating identification efforts. This approach plays a particularly critical role in countries that lack national DNA databases. In recent years, the identification of single nucleotide polymorphisms (SNPs) associated with facial morphology, combined with the widespread use of genome-wide association studies (GWAS), has made it possible to predict phenotypic traits related to external appearance directly from DNA. The present study evaluated the associations among 21 SNP markers (rs12633011, rs72691108, rs6555969, rs79909949, rs16863422, rs974448, rs1229119, rs12694574, rs805722, rs17447439, rs970797, rs12909111, rs2817419, rs7560738, rs619686, rs35965172, rs148037459, rs3736712, rs13267109, rs1074265, and rs507217) linked to eye morphology (including subtraits such as eye shape, upward or downward slant, eyelid configuration, and outercanthal width) and forehead structure (flat, posteriorly sloped, and pronounced posterior retraction). From the SNPs previously reported in the literature as phenotype-associated, those with the lowest p-values were selected, and a custom prediction panel was designed accordingly. Following optimization and validation, the panel was applied to volunteer samples from the Turkish population. DNA analysis was performed using the SNaPshot minisequencing method, and the resulting products were analyzed via capillary electrophoresis. Primers were initially optimized in singleplex and subsequently in multiplex panel. Validation experiments demonstrated that both the Mix-1 and Mix-2 panels produced complete genotype profiles at a DNA concentration of 1 ng/µl. Genotype–phenotype associations were assessed using logistic regression, Random Forest, XGBoost, K-nearest neighbors (KNN), and support vector classifier (SVC) algorithms. The models were evaluated through 5-fold cross-validation, and β-coefficients, odds ratios and model performance metrics (area under the curve [AUC] and receiver operating characteristic [ROC] curves) were calculated. Statistical analyses indicated significant associations between the rs507217 marker and En_Chr, the rs1229119 marker and both EnR_EnL and the upturned / downturned palpebral fissure, the rs7560738 marker and ExR_EnR, and the rs805722 marker and forehead morphology. Additionally, the rs2817419, rs6555969, rs35965172, and rs13267109 SNPs were found to have significant effects on intercanthal width, palpebral fissure, and overall forehead morphology. The findings suggest that the developed panel serves as an informative and effective tool for forensic DNA phenotyping (FDP) in the prediction of eye and forehead morphology. In conclusion, this research constitutes a preliminary study in facial morphology prediction and highlights the need for future investigations involving a greater number of volunteers and more diverse ethnic populations to enhance predictive accuracy and generalizability.