rs2378249

This study demonstrates that human hair color can be predicted from DNA variants with high accuracy using a multinomial logistic regression model. A subset of 13 genetic markers from 11 genes (MC1R, HERC2, IRF4, TYR, EXOC2, SLC45A2, TYRP1, OCA2, SLC24A4, KITLG, ASIP) predicted hair color categories in Polish Europeans with AUC values of 0.93 for red hair, 0.87 for black hair, 0.82 for brown hair, and 0.81 for blond hair. MC1R variants showed the strongest association with red hair (OR=12.64 for R variants, P=2.5×10⁻¹⁷), while rs12913832 in HERC2 was significantly associated with darker hair colors (OR=3.33 for black, P=4.3×10⁻⁶).