DNA Phenotyping Explained: How Police Predict a Suspect's Face from DNA

What the Method Measures

DNA phenotyping is a prediction system, not an identification system. The method starts from the biological fact that a defined set of genetic variants influences visible physical traits in statistically predictable ways. Eye color is driven primarily by variants in the HERC2 and OCA2 genes on chromosome 15. Hair color is driven by MC1R, ASIP, and a handful of related pigmentation loci. Skin pigmentation distributes across more than a dozen genes including SLC24A5 and TYR. Biogeographic ancestry is assessed through admixture analysis against reference populations. A sufficient DNA sample allows a prediction for each of these traits with a quantified confidence interval.

The dominant commercial implementation is Parabon NanoLabs’ Snapshot service, launched in 2015, which combines trait predictions with a rendered composite face. The HIrisPlex-S system, developed by Manfred Kayser’s group at Erasmus MC Rotterdam, is the open-access scientific standard for eye, hair, and skin color prediction and reports accuracy rates above 90 percent for the primary pigmentation categories in European-ancestry populations. Accuracy degrades somewhat in admixed populations and in underrepresented ancestry groups, reflecting the limits of the reference datasets on which the models were trained.

Where the Method Holds and Where It Breaks

The scientifically durable components of DNA phenotyping are pigmentation prediction and biogeographic ancestry. Both are supported by decades of population genetics research, are reproducible across laboratories, and produce predictions with quantified uncertainty. A Snapshot report that predicts brown eyes at 93 percent confidence and northern European ancestry at 86 percent confidence is communicating real information that a competent investigator can use to narrow a suspect pool.

The weak component is face prediction. Facial morphology is highly polygenic, meaning it is influenced by a very large number of genes each contributing small effects, and it is also shaped by non-genetic factors including nutrition, age, weight, and injury history. The face composites that Parabon produces are generated by combining pigmentation predictions with a generic face morphology that is adjusted slightly by a handful of shape-associated SNPs. The resulting images are not portraits of the suspect. They are visualizations of a prediction with substantial uncertainty that critics including the Electronic Frontier Foundation and several academic geneticists have argued can mislead investigators and the public.

The Cases Phenotyping Has Moved

The clearest successes for DNA phenotyping are cases where a Snapshot composite narrowed a suspect pool sufficiently to allow conventional or genealogical methods to close the identification. In the 1992 murder of Christy Mirack in East Lampeter Township, Pennsylvania, a 2017 Snapshot prediction of a Black or biracial male in his mid-twenties at the time of the crime was combined with a GEDmatch genealogy search in 2018 to identify Raymond Rowe. In the Ramsey Street Rapist case in Fayetteville, North Carolina, Snapshot predictions helped investigators focus on Darold Bowden, who was subsequently identified through additional DNA testing.

More commonly, phenotyping is used in tandem with investigative genetic genealogy rather than as an independent identification path. The Golden State Killer case did not rely on phenotyping because the genealogy search produced a direct family tree result, but the broader methodology of genetic genealogy investigations frequently uses phenotyping predictions to confirm candidates or prioritize branches during the tree-building phase.

The Regulatory Gap

Unlike genetic genealogy, DNA phenotyping operates almost entirely outside formal regulatory frameworks. There are no federal guidelines governing when law enforcement may request a Snapshot composite, what disclaimers must accompany the release of such a composite to the public, or how predictions of ancestry should be communicated to avoid reinforcing racial profiling. Critics have documented cases in which Snapshot composites were released to the public and functioned as de facto wanted posters despite wide confidence intervals on the underlying predictions.

The forensic genetics community continues to publish on the accuracy limits of face prediction and the communication practices that surround composite release. The science that underlies pigmentation and ancestry prediction is strong and continues to improve. The gap between what the underlying science supports and what commercial composites suggest about suspect appearance is the durable problem the method has yet to resolve.