—
## How HIrisPlex-S Works — and Why It Matters for Ancient DNA
When the Natural History Museum and UCL research team examined the Cheddar Man genome for appearance-related predictions, they were not simply looking at the raw DNA sequence and reading off a skin color. The process was considerably more structured, and the framework it used has its own history, capabilities, and limitations that bear directly on how the predictions should be interpreted.
The forensic prediction of eye color from DNA began as a practical need in criminal investigations. When biological evidence — blood, hair, saliva — is recovered from a crime scene and the contributor cannot be immediately identified through a database match, forensic biologists face a challenge: can they provide investigators with physical descriptions of an unknown contributor based solely on the DNA? The answer, for eye color, turned out to be yes, within limits.
The IrisPlex system, developed by a team including Susan Walsh, Peter Claes, and Manfred Kayser and published around 2011, showed that six SNPs — single nucleotide polymorphisms, positions in the genome where individuals commonly differ by a single DNA letter — could predict blue versus brown eye color with relatively high accuracy. The key variants were concentrated in and around the HERC2 and OCA2 genes on chromosome 15, which are strongly associated with iris pigmentation.
HIrisPlex extended the model to add hair color prediction. HIrisPlex-S, the most recent iteration published by Walsh and colleagues in 2017, added skin color to the prediction framework. Skin color prediction is substantially more complex than eye color prediction: it involves more genes, more variants, and a biological system that is less binary and more continuously variable than the relatively discrete blue-brown eye color distinction. The HIrisPlex-S system uses a model trained on modern reference populations to assign individuals to categories on a skin pigmentation scale.