U.S. flag

An official website of the United States government, Department of Justice.

NCJRS Virtual Library

The Virtual Library houses over 235,000 criminal justice resources, including all known OJP works.
Click here to search the NCJRS Virtual Library

Investigating the impact of age-depended hair colour darkening during childhood on DNA-based hair colour prediction with the HIrisPlex system

NCJ Number
Forensic Science International - Genetics Volume: 36 Dated: September 2018 Pages: 26-33
Magdalena Kukla-Bartoszek; Ewelina Pośpiech; Magdalena Spólnicka; Joanna Karłowska-Pik; Dominik Strapagiel; Elżbieta Żądzińska; Iwona Rosset; Marta Sobalska-Kwapis; Marcin Słomka; Susan Walsh; Manfred Kayser; Aneta Sitek; Wojciech Branicki
Date Published
September 2018
8 pages

This article presents research into predictive DNA analysis of visible characteristics and the impacts on forensic and anthropological investigations.


Predictive DNA analysis of externally visible characteristics exerts an increasing influence on contemporary forensic and anthropological investigations, with pigmentation traits currently being the most advanced for predictive modelling. Since pigmentation prediction error in some cases may be due to the result of age-related hair color darkening, and sex influence in eye color, this study aims to investigate these less explored phenomena on a group of juvenile individuals. Pigmentation phenotypes of children between the age of 6–13 years old were evaluated, in addition to data about their hair color during early childhood from a select number of these individuals. The HIrisPlex models for DNA-based eye and hair color prediction were used with input from SNP genotyping using massive parallel sequencing. Analysis of the total group of 476 children showed high accuracy in blue and brown eye color prediction, while hair color was predicted with AUC = 0.64 for blond, AUC = 0.64 for brown and AUC = 0.97 for red. 70.8% (n = 143) of the total number of children phenotypically blond for hair color during early childhood progressed to brown during advanced childhood. In 70.6% (n = 101) of those cases, an incorrect blond hair prediction was made during the time of analysis. A noticeable decline in AUC values for blond (from 0.76 to 0.65) and brown (from 0.72 to 0.64) were observed when comparing hair color prediction outcomes for the phenotypes recorded for the two different time points (at the age of 2–3 and 6–13). The number of incorrect blond hair color predictions was significantly higher in children with brown hair at age 6–13 who were blond at early childhood (n = 47, 32.9%), relative to children who had brown hair at both time points (n = 6, 9.4%). However, in 28.0% (n = 40) of children who did experience hair color darkening, HIrisPlex provided the correct prediction for the darkened hair color phenotype, despite them being blond in early childhood. The authors’ study implies that HIrisPlex can correctly predict adult hair color in some individuals who experience age-related hair color darkening during adolescence. However, in most instances prediction seems to default to the pre-adolescent hair color for individuals with this phenomenon. In the future, the full adolescent age range in which hair color darkening can occur should be considered in the study samples used for training hair color prediction models to obtain a more complete picture of the phenomenon and its impact on DNA-based hair color prediction in adults. (Published Abstract Provided)