DNA analysis is less accurate for certain groups of people 

That swab of DNA from a door handle or drinking glass isn’t always a foolproof way to determine who committed a crime.  

Forensic DNA analysis is less certain when untangling DNA mixtures from groups with low genetic diversity, University of Oregon researchers reported this fall in the journal iScience. In those cases, the technique is more likely to incorrectly link an innocent person to the scene of the crime — with potentially life-changing consequences. 

Today’s modern DNA analysis techniques are extremely sensitive. Instead of relying on a drop of blood, forensic scientists now can gather trace amounts of DNA left behind by shed skin cells and match that DNA to a particular person. However, that also means scientists are often analyzing mixtures containing DNA from many different people, like everyone in the home who has recently touched the door handle, for example.  

By looking at the variation of certain genetic markers in the sample and comparing it to the suspect’s DNA, researchers can quantify the strength of the evidence linking a particular person to the mixture.

DNA mixture analysis can be powerful when used correctly, but it’s crucial to understand the technique’s limitations and when it should be wielded with particular caution, said Rori Rohlfs, a data scientist at the University of Oregon who led the work alongside a group of undergraduate researchers from San Francisco State University. 

Rohlfs and her team wanted to see how the accuracy of that approach is affected by people’s genetic ancestry. They combed through previously published genetic databases to get data on the frequency of certain genetic variants for groups of people with different genetic ancestry. Then, they used forensic analysis software to simulate profiles of individuals, as well as mixtures of DNA representing groups of people from different genetic backgrounds.

In mixtures with lower genetic diversity, the team found the technique was more likely to yield a false positive; that is, to incorrectly link someone to the mixture who wasn’t actually involved. And the problem worsened when the mixture contained DNA from more people. 

"The accuracy of DNA mixture analysis really varies by genetic ancestry,” Rohlfs said. “Groups with less diverse genetic variants are going to have higher false inclusion rates for DNA mixture analysis, and this gets worse when you have more contributors.” 

The study involved simulated genetic mixtures generated from complex datasets, so the researchers can’t specifically say that the technique is less accurate for certain genetic ancestry groups. Plus, traditional race and ethnicity labels are often overly broad and don’t always accurately map to genetic ancestry, Rohlfs said.

“The accuracy of DNA mixture analysis really varies by genetic ancestry. Groups with less diverse genetic variants are going to have higher false inclusion rates for DNA mixture analysis, and this gets worse when you have more contributors.” 

Rori Rohlfs, UO data scientist

But some examples of groups that might have less genetic diversity include certain Indigenous, Latine or Pacific Islander groups. 

The research also reflects the challenges of doing ethical genetic research, Rohlfs said. Many of the studies her team looked to for data did not necessarily practice informed consent when collecting people’s DNA, sometimes getting samples from incarcerated people. The research team opted to only include data from subjects where informed consent could be verified, somewhat limiting their data pool.

Rohlfs is currently building up her lab at the UO, and she hopes to continue investigating the accuracy of other emerging forensic DNA analysis techniques.