The young couple set out on a trip in 1987, speeding toward Seattle, Washington, in a gold van, when they crossed paths with a killer.
The man raped Tanya Van Cuylenborg and shot her in the head. Jay Cook was beaten and strangled.
The killer left a pair of plastic gloves inside their vehicle, a gesture one detective interpreted as a taunt: You'll never catch me.
That was true for more than three decades.
Investigators spent thousands of hours sifting leads and probing suspects with little to show. But in late April, a former musical stage actor with no background in law enforcement took over the case.
CeCe Moore and her team cracked it in three days.
Moore put the killer's DNA profile into a public genealogy website to find relatives and then built a family tree that led to a suspect, William Earl Talbott. The truck driver was charged in Washington state in May.
Since the same technique was used to find the suspected Golden State Killer in April, genetic genealogy has led to a flurry of breakthroughs in the coldest of cases, showing the potential to be a transformative tool for police.
Yesterday, police in Indiana announced that Moore's team at Reston, Virginia-based Parabon NanoLabs had helped identify a man who allegedly sexually assaulted and killed 8-year-old April Tinsley in 1988.
The killer had sent chilling messages to others over the years, tacking some of threats on bicycles belonging to other young girls.
Parabon, the biggest player so far to work in the emerging field, also helped authorities identify a Pennsylvania DJ charged with the 1992 slaying of a primary school teacher, the prime suspect in the 1981 killing of a real estate agent in Texas, and a Washington man charged in the 1986 rape and killing of a 12-year-old girl.
Another team is working with investigators in California to try to solve the Zodiac Killer case.
And the nonprofit DNA Doe Project has uncovered the identities of a man who mysteriously assumed an 8-year-old's identity before committing suicide, a woman slain in Ohio in 1981 and two others as it works to put names to the remains of 40,000 Jane and John Does scattered across the country.
The developments are all the more remarkable because genetic genealogy was not pioneered by the FBI or elite forensic experts but by a loose network of citizen scientists and genealogists like Moore and a professional guardian from Florida, who came up with the idea for a genealogy database available for all to search.
But the novel turn to crime-fighting has raised a host of issues: Could the technique finger the wrong person? Who will ensure police use genetic data responsibly? Should authorities rely on a public database that could be hacked or manipulated?
"We've got no precedent for doing this type of thing," said Debbie Kennett, a research associate in the Department of Genetics, Evolution and Environment at University College London. "The people who are doing it are often volunteers."
The hunt for Van Cuylenborg and Cook's killer began with the genetic equivalent of a Google search on a Friday night.
Moore's team took a profile of the killer's DNA obtained from the crime scene and provided by authorities and uploaded it to GEDmatch, a genetic clearinghouse that allows users to find relatives by comparing their genetic code against more than 1 million others.
GEDmatch's analysis represents a quantum leap over traditional DNA matching used by law enforcement since the 1980s. In those cases, a lab takes a sample that contains up to 20 short segments of the perpetrator's genetic code and looks for a match in a state DNA database or the FBI's Combined DNA Index System, which contains 17.3 million profiles.
The profiles uploaded to GEDmatch contain some 600,000 DNA snippets, allowing the genetic genealogist to not only identify a match but also to determine how closely people are related.
GEDmatch spat out results for Moore after roughly eight hours of crunching data: the killer appeared to share enough DNA with two people to be second cousins.
Moore was excited. She knew the relatives came from different branches of a family because they did not match each other genetically. The key would be to find where their lines intersected.
She traced both back to great-grandparents and then began creating a family tree flowing forward in time.
Moore scoured census records, death and marriage records, newspaper archives, social media, and other sources to find descendants. Over the weekend, she came across a newspaper obituary that stopped her cold.
The clipping was for a member of one branch of the suspect's family tree, and it mentioned Talbott's mother, who also had a surname from the other branch. Moore had found the trunk where the branches met.
"That is the eureka moment," Moore said. "These are two unrelated people who share significant portions of DNA with the suspect, and then there's a marriage between their families."
It was a short leap to Talbott, who lived near the scene of the killings at the time.
Most investigations are not so quick or straightforward. In some cases, the search can only be narrowed to a handful of suspects or even just a country where the suspect's family originated and possible surnames.
In other cases, GEDmatch may find relatives so distant it is difficult to track them to a potential suspect, or efforts to piece together a family tree may hit a dead end. Parabon, hired by law enforcement agencies across the country, performs an initial assessment on a case for US$1500 and then charges US$3500 to work the genetic genealogy, which is turned around in 45 business days or less.
Moore turned Talbott's name over to authorities. Police in Washington scooped up a cup that he discarded and performed a DNA test on some genetic material on it. Police said it was a match with DNA from the crime scene.
Talbott was arrested on May 17.
John Van Cuylenborg said genetic genealogy represents a "ray of hope" for families like his. He said he had resigned himself to the unsettling likelihood his sister's killer would never be caught. He was whipsawed by the speed with which genetic genealogy led to an arrest.
"Suddenly, somebody was able to be held to account," Cuylenborg said.
Genetic genealogy was kick-started around 2000 when a company called FamilyTreeDNA began offering consumer DNA tests that allowed users a new tool to explore their heritage.
But the screenings only allowed users to see slivers of their paternal and maternal lines until 2009, when 23andMe introduced an autosomal DNA test that allowed users to find relatives throughout a family tree.
Moore said she saw the potential and dropped her entertainment career. She tested 40 of her own family members and began blogging on what she was finding. She had no science degree but said she was getting readers from Harvard, Stanford and other universities.
She and others had started out simply trying to find distant ancestors with genetic genealogy, but she was soon receiving requests from adoptees: Could you help me find my biological parents?
It was an intriguing puzzle. Moore knew "search angels" who were helping adoptees through records searches. She figured she could meld her genetic approach to their detective work.
She refined her methods and began helping adoptees reconnect with their biological families.
The work grabbed the attention of Harvard University professor Henry Louis Gates, who added her in 2013 as a genetic genealogist to his PBS show, Finding Your Roots. In the years that followed, Moore's work also caught the attention of detectives and forensic experts, who began asking whether genetic genealogy could help law enforcement.
"Those techniques were the basis of what we are now doing," Moore said.
Then last year, genetic genealogist Colleen Fitzpatrick set out to bring answers to families whose loved ones had gone missing. With partner Margaret Press, she started the DNA Doe Project, a group of genealogists who volunteer to put names to unidentified human remains.
One was a woman with red braids who was found dead in a ditch in Troy, Ohio, in 1981. She had been strangled and beaten but had nothing on her that gave a clue to her identity.
Police publicly released sketches, but nobody was able to identify her. She became known by the distinctive jacket she wore - "the Buckskin Girl."
She was eventually buried under a headstone that read simply: "Jane Doe."
The criminal investigation stalled, and decades passed.
Then a forensic anthropologist on the case enlisted the DNA Doe Project's help. They fed the Buckskin Girl's genetic profile into GEDmatch and were monitoring the results when a match was made with a first cousin once removed. It electrified the group.
"That's almost like shaking hands with the person for a genealogist," Fitzpatrick said.
Fitzpatrick and Press discovered the person had uploaded a family tree on Ancestry.com. As they explored it, Press suddenly went silent.
"'Oh, my God. You won't believe this,'" Press told Fitzpatrick.
Press was looking at an entry for a relative in Arkansas named Marcia King. The page listed her birth date as 1959, but under her date of death it said: "Missing - assumed dead."
After nearly 40 years, "The Buckskin Girl" had a name again.
It was a short leap from cases like King's to solving crimes with genetic genealogy, but both Fitzpatrick and Moore were reluctant. The idea of helping law enforcement trawl DNA databases without users' knowledge made them uneasy.
But even as they grappled with these concerns, an effort was underway in California to do that very thing. A soon-to-retire investigator thought genetic genealogy could provide one last shot at cracking his biggest unsolved case: the Golden State Killer.
Paul Holes worked with a genetic genealogist who has remained anonymous to upload the killer's DNA profile to GEDmatch late last year. Genealogy websites such as Ancestry.com and 23andMe had millions more profiles, but unlike them, GEDmatch had no policy against searches by law enforcement.
Holes' team found distant relatives, which allowed them to build a family tree and eventually identity and arrest Joseph James DeAngelo, a retiree who was living quietly outside of Sacramento. He's been charged in 12 slayings.
Curtis Rogers, one of GEDmatch's founders, said he was as surprised as anyone that GEDmatch had played a leading role in identifying a suspect in one of the nation's worst serial slayings and is now becoming a major resource for law enforcement.
The site, created in 2010, is a passion project for the professional guardian from Florida and a couple of others. It has no full-time staff and has the bare-bones look of a page from the dawn of the Internet.
"If we had good tools, we wanted to share them," Rogers said of GEDmatch.
It was initially only a destination for genealogy buffs. But each year, Rogers said the number of profiles in the database roughly doubled.
The surreptitious use of the site by Holes and his team turned Moore's private qualms about using genetic genealogy to solve crimes into a public firestorm. Some complained GEDMatch had not warned users the site might be used by law enforcement. Privacy advocates saw the case as a cautionary tale about sharing sensitive personal data online.
Rogers said he had no idea Holes and his team were using GEDmatch, but he has come to support such efforts after much deliberation. He recently updated GEDmatch's policy to make it explicit that police may use the database, ensuring genetic genealogy is here to stay as a law enforcement tool.
He said he was touched when the daughter of a suspected serial killer asked him to include her DNA profile in GEDmatch to potentially help police solve cases.
Moore and Fitzpatrick said they now feel comfortable working with law enforcement because the publicity surrounding the Golden State Killer case made it clear to GEDmatch users that police might tap the database.
In May, Moore announced her partnership with Parabon NanoLabs. Steve Armentrout, Parabon's chief executive, said the future of genetic genealogy is using it at the start of an investigation, not just in cold cases. They have already taken that step in some.
Detective Chris Flanagan, of the Fairfax County police cold-case unit, said that innovation could be particularly helpful. "If you were able to limit the suspect pool . . . you could focus resources and greatly speed up the investigation," he said.
But the rapid adoption of genetic genealogy by law enforcement has given others pause. Kennett, the research associate at University College London, said there is no oversight for how police use GEDmatch, as there would be with a law enforcement database. Nor are there best practices for genetic genealogists conducting searches or anyone certifying their skills.
"There have been cases in the adoption community where people have been reunited with the wrong parents because of misinterpretation of data," Kennett said. "If that can happen in an adoption search, it could also happen in a criminal search, with much more adverse consequences."
Moore said she has never made an incorrect match but shares some of Kennett's concerns. She is working on best-practice guidelines for genetic genealogists and points out that the work is just a tool. Any identification is confirmed through a DNA test by police before an arrest is made.
"People will have to think twice about committing these type of crimes because it's going to be a lot easier to identify them," Moore said.