Forensic DNA Phenotyping of Externally Visible Characteristics and Biogeographic Ancestry in Indonesian Populations: A Cross-Sectional Validation Study

Abstract

Background. Forensic DNA phenotyping (FDP) infers externally visible characteristics (EVCs) and biogeographic ancestry (BGA) from DNA when no reference profile is available. How well it performs in Indonesian populations has been little characterised.

Objective. To evaluate how accurately a single nucleotide polymorphism (SNP) panel predicts skin, iris and hair colour and broad regional ancestry in Indonesian adults.

Methods. In this cross-sectional study, 1,000 unrelated adults from eight named ethnic groups and a composite group were genotyped at 20 pigmentation SNPs and 26 ancestry-informative markers by multiplex single-base extension. Colour categories were predicted with a multinomial logistic model calibrated to HIrisPlex-S, and ancestry zones (Western, Central/other, Eastern) with a random forest. Accuracy was compared with the no-information rate, the accuracy of always predicting the commonest class.

Results. Skin colour was predicted correctly in 89.2% of participants (no-information rate 45.0%), iris colour in 82.0% (88.0%) and hair colour in 80.0% (72.0%). The area under the curve was 0.96 for dark skin, 0.88 for brown irises and 0.87 for black hair; sensitivity for non-brown irises was 52.5%. Overall, 83.7% of participants were assigned to their expected ancestry zone (no-information rate 66.5%); sensitivity for the Eastern zone was 80.0% (95% CI 60.9–91.1; n = 25).

Conclusion. In this cohort, skin colour prediction clearly exceeded the baseline, whereas iris and hair colour prediction improved little on it. The panel may support investigative leads, but independent test samples are needed before casework use.