In recent months, several murder suspects were caught after detectives matched DNA samples with information on genealogy websites, locating them through their family trees

The Golden State Killer evaded the police for over forty years. Between 1974 and 1986, he raped over fifty women, and according to police estimates, murdered twelve people at different locations throughout California. Armed with a gun, his face covered with a ski mask, he broke in to houses. He tied his victims and took mementos – pictures and jewelry. Sometimes he even paused to eat and drink in the houses he broke into, and it seemed he thought no one could catch him.

And he had good reason to think so: For decades, police detectives failed to connect the crimes committed at different locations throughout the state, with each police department investigating the events under its authority separately. It was the DNA evidence from the rape and murder scenes that eventually revealed it was a single perpetrator in all of them, but the investigation still did not advance significantly.

In April 2018, Joseph James DeAngelo, 72, a former police officer, was arrested; the Golden State Killer was finally caught. The breakthrough in the investigation came thanks to the creative use of DNA samples at the police's disposal: Law enforcement officials uploaded them to a genealogy website that connects people with genetic relations, and thus found DeAngelo's relatives. From there, the road to arrest was short.

And he was not the last. Within a month, more than a hundred DNA profiles were uploaded to GEDmatch, the website used in the Golden State Killer investigation. These efforts have already led to the arrest of murder suspects in Pennsylvania, Indiana, and the Washington State. Texas officials notified the police that they have found new evidence in a murder case committed 37 years ago. How does the police use these websites and the information they store? And what repercussions does this have for our families and our privacy?

Collecting a blood sample | Photograph: Shutterstock
DNA samples from crime scenes can be later compared to online databases. Collecting a blood sample | Photograph: Shutterstock

Online family tree

Companies that perform private genetic tests and give their customers information regarding their family tree have become very popular in recent years. The company 23andMe has about five million customers, Ancestry.com has about ten million, and the Israeli-based company MyHeritage has about a million users. If you send any one of these companies a spit sample, along with payment, of course, they will supply you with your genealogical information, including ancestry – such as where your ancestors came from, your relation to different ethnic groups, and even the amount of Neanderthal DNA in your genome. In addition, they allow you to search for relatives that are also customers of the same company, according to the genetic match between you.

GEDmatch does not perform genetic testing itself. It allows users to upload genetic information acquired from another company and compare it to information on its database, coming from other users. This enables users to find relatives that also use the website, regardless of the company which did the actual genetic testing, and then build an elaborate family tree. Over a million people have already uploaded their genetic profile to the database.

DNA test | Photograph: Shutterstock
Millions throughout the world have already uploaded their DNA sequences to online databases. DNA test | Photograph: Shutterstock

All in the extended family

This treasure-trove of genetic information is now available for detectives looking for rapists and murderers. For instance, in the 1987 murder of the couple Tanya Van Cuylenborg and Jay Cook near Seattle, WA, the murderer left a pair of plastic gloves at the crime scene, from which the police was able to produce his DNA. Nevertheless, for three long decades, it did not bring them any closer to the murderer himself.

In May 2018, genetic genealogist CeCe Moore of Parabon NanoLabs joined the investigation. Moore and her team uploaded the genetic profile from the sample to GEDmatch, and after about eight hours, found two individuals that were, according to DNA compatibility with that of the murderer's, his second-degree cousins. Since these two people were not related to each other, the investigators concluded that they were cousins from different sides of the murderer's family. Moore realized that if she could find where the meeting point of these two families, she would find the person she is looking for.

For this purpose, the investigators acquired census data, marriage, birth, and death records, and also browsed archives of newspapers and social networks. They began constructing the family trees of both second-degree cousins and soon enough, found the one piece of information that was the key to solving the case. In a eulogy published in the newspaper following the death of a member of one of the families, a relative with the last name of the other family was mentioned.

“That is the eureka moment,” Moore said to The Washington Post. “These are two unrelated people who share significant portions of DNA with the suspect, and then there's a marriage between their families.” A quick investigation found that the same woman had a son of the right age that lived not far from the place where the murder was committed, and Moore gave his name to the police. This entire process took only three days.

The police followed the suspect and collected a DNA sample from a cup he had thrown away. A test confirmed the DNA from the cup matched the murder scene sample. This is how William Earl Talbott, 55, was caught 30 years after the act.

 The detectives did not connect between the murders he committed. Facial composites from the beginning of the chase after DeAngelo | Source: Wikipedia
The detectives did not connect between the murders he committed. Facial composites from the beginning of the chase after DeAngelo | Source: Wikipedia 

Deep database

Not all cases were as quick and easy to solve: Sometimes, GEDmatch can locate only more distant relatives, or family tree reconstruction proves a more complex task. Nevertheless, the website definitely provides a new, at times a much more efficient, way of identifying suspects. What is so unique about this website that leads police and FBI detectives to turn specifically to it – and not to the FBI's DNA database?

In the 1990s, the FBI founded the Combined DNA Index System (CODIS). To date, it contains genetic information from over 17 million people, far more than GEDmatch. However, the CODIS information is different from the profiles on the website, which come from 23andMe or Ancestry.com. The main objective of CODIS is to check whether the database contains a DNA profile that matches crime scene samples, not to find relatives. Therefore, the investigation focuses on regions of our genome with high variance among individuals: Around twenty such regions are sufficient for determining if a blood drop found at the crime scene belongs to the suspect with very high certainty. Such knowledge is not sufficient, however, to tell us nearly anything else, and cannot indicate relatives beyond first-degree.

The surveys performed by the private companies are much more comprehensive and include identification of the “letters” – the DNA building blocks – at about 600,000 locations throughout the genome. This allows them to identify variations in genes that are related to different traits, to compare different genomes, and determine whether their owners are first, second. or even third-degree relatives. Today it is possible to perform these tests relatively quickly and they cost less than in the past, yet are still more expensive than the tests conducted at forensic laboratories. During most of CODIS' existence, performing such comprehensive tests was a very complicated process, beyond the FBI's budget.

This is what led police and FBI detectives to GEDmatch. But not everyone is certain it is such a good idea.

I never expected anything like this,” admitted Curtis Rogers, who started GEDmatch along with John Olson to The Atlantic. The two founded the company as a side project, which was not their primary occupation. Most of the website’s options are free, and its design reveals that neither a lot of money nor a large company was involved in its establishment. Nevertheless, it has become the first choice for genealogists.

Privacy policy

Rogers and Olson never thought their side project would become so popular with law enforcement authorities, and did not know that police detectives used it to search for the Golden State Killer, until his arrest was publicized. “I didn’t like this use of our website,” said Rogers.

The website is public and anyone can use it – though in the past, users had to declare that the genetic information they upload is in fact their own, belongs to someone they are the legal guardians of, or that they were “authorized” to use this information (it is not exactly clear what this kind of authorization refers to). It is not clear how uploading a DNA sample from a crime scene fulfills these conditions, and law professor Erin Murphy told The New York Times that this might raise questions about the evidence’s admissibility.

Following the Golden State Killer’s arrest, Rogers and Olson changed their website's terms of service and privacy policy, which now also includes uploading crime-scene DNA samples by the authorities for the purpose of identifying a perpetrator of a “violent crime,” such as rape or murder. They informed users that law enforcement officials are using the information on the website, which led some to leave it and delete their genetic profiles. But many others, says Rogers, sent him thank-you letters.

A family tree | Illustration: Shutterstock
Using genetic information not to investigate customers, but rather, their relatives. A family tree | Illustration: Shutterstock

Participating in an experiment

"What is the right thing to do? I’m still not sure,” he says. “I think the best thing we can do is admit what we are open and honest.”

The ethical problem stems, in part, from the fact that the investigation uses genetic information that website users upload not to interrogate them, but to reach their relatives, which, obviously, never gave their consent to share such information with the public, or with the police.

“Suppose you are worried about genetic privacy,” Murphy said. “If your sibling or parent or child engaged in this activity online, they are compromising your family for generations.”

Website customers are also divided on this issue. On a Facebook page dedicated to genealogy, one user responded with outrage: “My relatives consented for their data to be used for genealogy but not for criminal investigations,” However, another user wrote: “I’ll volunteer to give my DNA and out any of my cousins who may be rapist/murderers. So much drama over nothing,”

Nowadays, police detectives are, in fact, using the website to track down murderers, but are we certain they will not expand this usage in the future? Blaine Bettinger, a lawyer affiliated with GEDmatch told The Atlantic that he is concerned that the police will begin searching for offenders of less serious crimes, like drug offenses.

“That makes me uncomfortable," he says. "We leave DNA everywhere we go. Everywhere we touch has DNA. There’s got to be a limit.” Though GEDmatch’s terms of service demand that uploaded DNA samples from crime scenes will only come from murder and rape cases, the site's owners do not have any way of verifying this. Generally speaking, since the website is public and does not belong to any government authority, it is not subject to any sort of restrictions or supervision. In contrast, some states in the US have laws that limit CODIS use, for example.  

Police detectives' abrupt entry into the genetic genealogy field has left legislators behind: Attempts to supervise such investigations will surely follow, but until they do, it seems anything goes. “People on GEDmatch are part of an experiment,” concluded genealogist Debbie Kennett in the interview for The Atlantic. An experiment that they did not necessarily sign up for – but no one asked them.

 

Translated by Elee Shimshoni