All in the Family: Finding Criminals Through the DNA of their Relatives

The recent arrest of Joseph James DeAngelo Jr., the suspect in the Golden State Killer investigation, is remarkable for the striking success of finding this suspect by searching a genealogic database. Use of public genealogy databases for forensic purposes is not yet widespread and some may be concerned about such uses. Even in light of recent privacy breaches involving companies like Facebook and Cambridge Analytica, Californian authorities did the right thing here in the interests of justice and public safety and, against the odds, hit the jackpot.

Searching a DNA profile against a publicly or privately held genealogical database will typically yield a very large number of possible biological relatives who may be widely dispersed across the globe. This is because these genealogic and ancestry databases contain profiles from tens of thousands of volunteers and test for types of DNA variation known as SNPs. They also sequence maternally inherited DNA along with Y-chromosome-specific DNA in males. Many individuals, who may be quite distant relatives, are thereby identified. It is quite remarkable, almost stunning, that searching a genealogic database was successful in the Golden State killer case, as it was only by chance that someone related to DeAngelo had voluntarily uploaded their own DNA profile to a public online website to try to find their relatives. Doing the same thing with the DNA profile of biological evidence from one of the Golden State Killer crimes, Californian authorities eventually sorted through the list of possible relatives and eventually focused their investigation on DeAngelo.

Several web sites allow anyone to perform such searches on their own DNA or that of those for whom they have guardianship. Once a list of possible relatives is obtained, law enforcement could easily use “meta-data,” such as geography and age, to focus attention on specific individuals most likely living near the crime scenes who have certain relevant characteristics. For example in the California case, as the crimes occurred sometime in the 1970s or '80s, the perpetrator must have been at least 50 years old. DNA technology is now rapidly evolving to predict many physical characteristics about someone simply from analysis of a DNA sample. New tools and technologies now developed can be used for both humanitarian purposes and for solving crimes and, accordingly, public officials will undoubtedly be asked to look closely at their use.

While concerns may loom large for some about data mining methods it is important to understand that the arrest in the Golden State killer case was not accomplished by the more focused protocols using the DNA kinship searching methods that my colleagues and I described over a decade ago. Such kinship searching methods, known popularly as familial searching, involve a multi-step search of local or state databases containing only DNA profiles of convicted offenders, collected under state law. In this approach the offender DNA profiles are ranked in decreasing likelihood of being derived from a close relative of the crime scene DNA. The DNA from a small number of closely matching profiles, along with the crime scene evidence, is then subjected to further testing, known as Y-STR testing, to winnow out non-relatives. When successful, a familial search yields a single DNA profile. The name of the convicted offender associated with that DNA profile can then be released to authorities for further investigation, in order to determine whether a close relative could be a viable suspect in the particular investigation. Because it is quite common for criminals to have a relative whose DNA profile in the offender database, such familial searching methods have now led to arrests and convictions in several U.S. states and other countries, including success on very first attempts in several states including California and Arizona.

While the California Department of Justice did attempt this familial searching method several times in this Golden State killer case, each without success, overall these more focused formal kinship searching methods are far more likely to resolve crimes, with minimal intrusion on privacy, than are searches of public datasets which now include hundreds of thousands of individuals. As we continue to freely give up our private information in so many ways clever government organizations and big-data companies may choose to leverage these data. These choices require us to weigh the delicate balance between competing interests of privacy and public safety.

Frederick R. Bieber is Associate Professor of Pathology at the Medical School and a published author and expert on forensic DNA.