BR-FDP-EYE: Brazilian Forensic DNA eye phenotyping

Abstract

Forensic DNA phenotyping (FDP) is a technique used to predict an individual’s physical appearance traits based on DNA. Eye color, a highly heritable genetic trait, is one of the most obvious and distinguishable externally visible characteristics (EVCs) employed in human identification. Genotype-phenotype association studies have focused on detecting single nucleotide polymorphisms (SNPs) located in and nearby genes directly or indirectly involved in pigment synthesis. In this study, we investigated 66 SNPs related to pigment production and their efficacy in predicting eye color for forensic purposes. Using bioinformatics and machine learning techniques, we analyzed these SNPs in 438 admixed individuals with diverse phenotypes from Southern Brazil. We introduce BR-FDP-EYE, a system for phenotype prediction from DNA. This eye color predictor utilizes five to seven SNPs to classify phenotypes, achieving an overall accuracy of up to 85%. Users can select from four classification methods. The first option is a five-class classifier (BR-FDP-EYE-5), distinguishing Blue, Green, Hazel, Light Brown, and Dark Brown eye colors. While this classifier demonstrates high sensitivity for the Blue and Dark Brown classes (81%–87%), it exhibits lower sensitivity for the remaining three classes (up to 33%). Alternatively, users can choose one of three three-class classifiers, depending on their specific needs: BR-FDP-EYE-3-I for Blue, Intermediate (Green ＋ Hazel ＋ Light Brown), and Dark Brown; BR-FDP-EYE-3-II for Light (Blue ＋ Green), Hazel, and Brown (Light Brown ＋ Dark Brown); and BR-FDP-EYE-3-III for Blue, Intermediate (Green ＋ Hazel), and Brown (Light Brown ＋ Dark Brown). These three-class classifiers improve overall accuracy and enhance per-class metrics, providing a more balanced performance for targeted applications. BR-FDP-EYE offers a simple, user-friendly interface suitable for genetics and anthropological research, educational purposes, and potentially, law enforcement investigations.