Genetic analysis of 22 pigmentation-associated SNPs in the Pashtun population of Khyber Pakhtunkhwa, Pakistan: Implications for forensic DNA phenotyping

Abstract

The prediction of externally visible characteristics (EVCs) from DNA, known as forensic DNA phenotyping (FDP), has emerged as a powerful tool in forensic investigations. The HIrisPlex system, comprising 24 SNPs for eye and hair color prediction, has been extensively validated in European populations but shows reduced accuracy in South Asian populations. This study genotyped 22 pigmentation-associated SNPs using the SNaPshot minisequencing platform in 146 unrelated Pashtun individuals from Khyber Pakhtunkhwa (KP), Pakistan, to evaluate their forensic utility and phenotypic association in this understudied population. Minor allele frequencies (MAFs) ranged from 0.010 (rs885479) to 0.394 (rs28777), with 13 of 22 SNPs (59.1%) being polymorphic. Nine SNPs (N29insA, rs11547464, rs1805008, rs1805005, rs1805006, rs1805007, rs1805009, Y152OCH, rs1110400) were monomorphic, consistent with the dark pigmentation characteristic of South Asian populations. Hardy-Weinberg equilibrium testing revealed four SNPs (rs885479, rs4959270, rs12203592, rs2378249) deviating from equilibrium at p < 0.05. Polymorphism information content (PIC) values ranged from 0.020 to 0.364 (mean = 0.128). Linkage disequilibrium analysis of 78 SNP pairs revealed no strong LD (r² > 0.8), confirming marker independence. Pairwise FST analysis among six major ethnic subgroups (Yousafzai, Afridi, Khalil, Khattak, Momand, Wazir) indicated low genetic differentiation (FST < 0.05 for 13 of 15 comparisons). SNP-phenotype association analysis identified rs12913832 (HERC2) as significantly associated with eye color (chi2 = 44.24, p = 3.02 × 10^-7, Cramer's V = 0.389) and rs16891982 (SLC45A2) as significantly associated with hair color (chi2 = 23.77, p = 0.008, Cramer's V = 0.285). Multinomial logistic regression models with 5-fold stratified cross validation achieved 88.3% accuracy for binary eye color (dark vs. light) and 83.6% for binary hair color, while multiclass predictions were lower (37.7% for eye color, 62.3% for hair color). These findings establish baseline population data for forensic DNA phenotyping in the Pashtun population and highlight the need for population-specific prediction models in South Asian groups.