• This study focused on analyzing the association between six SNPs located in pigmentation genes to the freckling phenotype in a sample of 534 Brazilians. • Binomial logistic regression approach was used to develop three prediction models: the DNA-based, the sex-adjusted and the ancestry-adjusted prediction models. • Our findings indicate that rs12203592 in IRF4, rs1800404 in OCA2 and rs2228479 in MC1R , along with the sex and ancestry variables, present great potential as predictors of freckles in the Brazilian population. The prediction of externally visible characteristics (EVCs) is widely used in the forensic sciences to assist investigations in which the identity of perpetrators or victims is unknown. Genetic markers associated with pigmentation traits have been extensively studied for predicting phenotypic traits such as hair, eyes and skin color, as well as freckles, through the analysis of single nucleotide polymorphisms (SNPs). This study analyzed the association between six SNPs in pigmentation genes (rs1042602, rs1800404, rs11636232, rs2238289, rs12203592, rs2228479) and the freckling phenotype in 534 adult Brazilians for forensic purposes. Results indicated that three polymorphisms present a significant association with the studied EVC (p < 0.05). To assess predictive performance of the tested SNP set, three prediction models were developed using binomial logistic regression. The first model included three genetic variants and presented moderate predictive performance, which was improved when sex was included as a predictor in the second model, reinforcing its importance for freckle prediction, as reported in previous studies. Biogeographical ancestry was included in the third model, which achieved the highest discriminative ability. ROC-derived optimal cut-off thresholds were estimated to find the best balance between sensitivity and specificity, indicating how these parameters can be strategically defined depending on the forensic application. Overall, the predictive performances obtained are consistent with those reported for previously published freckle prediction tools, supporting their potential value for forensic applications. Further analyses to refine and validate the models should be carried out to evaluate their applicability for forensic phenotyping in the Brazilian population.
Forlenza et al. (Sun,) studied this question.