The Cold Case Factory


April Tinsley’s murder should have been easy to solve. Three days after her April 1, 1988, disappearance, when the eight-year-old girl’s body was discovered in a ditch in Fort Wayne, Indiana, police were able to collect plenty of clues that could help identify the culprit. They found one of April’s shoes lying near her body, as well as a shopping bag containing a dildo. DNA could be extracted from this physical evidence, which made the absence of any leads to the murderer’s identity incredibly frustrating for the police. As long as the perpetrator’s DNA had no corresponding match among the information cataloged in CODIS, the FBI’s flagship DNA database, or in similar databases at the state level, the prospect of an arrest dimmed.

Early optimism gave way to anguish, frustration, and fury, exacerbated when notes started to appear a few years after April’s murder. One turned up in 1990, scrawled on a barn door: I kill 8 year old April M Tinsley. Then nothing until 2004, when four handwritten notes appeared in mailboxes across Fort Wayne, and plastic bags containing used condoms and obscene photos of a man’s penis were found on girls’ bicycles. The notes mocked April’s family and taunted the police, while also threatening to kidnap and kill more victims.

Forensic DNA testing wasn’t common when April Tinsley was murdered in 1988. It was cumbersome, time-consuming, expensive, and the tests required much more physical evidence—semen, blood, and saliva—than is common today. DNA tests would only become de rigueur in the mid-1990s, when the protocol changed and investigators could now amplify small amounts of DNA taken not just from bodily fluids, but from objects like guns, clothes, and tools.

The trouble was, even as Fort Wayne police kept up with advances in testing, the DNA they had from the scene where April’s body was found simply wasn’t enough to produce a conclusive result. By 2015, even though the sensitivity of DNA testing had improved enough to detect DNA at the level of a single nanogram—a grain of salt is about 58,000 nanograms—the evidence still didn’t match anyone in the FBI database.

In June 2015, Fort Wayne police learned that Parabon NanoLabs, a biotech company headquartered in Reston, Virginia, was offering a service called Snapshot, in which a working sketch of a criminal suspect could be generated directly from minuscule amounts of DNA. Parabon’s roots were in bioinformatics, and their techniques, including Snapshot, which was invented and trademarked by the company in 2012, were originally designed for use in medical research.

Snapshot uses a process known as DNA phenotyping to determine several distinct physical traits—eye color, hair color, nose shape—to produce computer-generated sketches of a suspect. The sketches of April Tinsley’s killer, generated in May 2016, showed what the suspect looked like in 1988 and what he might look like 30 years later: dark hair, hazel eyes, prominent nose, with gray flecks within his sideburns in the age progression. Despite the lifelike quality of the sketches, which appeared throughout Allen County on the local news and were distributed via the police department’s social media channels, they did not lead to an identification, or an arrest.

Two years later, in 2018, detectives in Fort Wayne learned that Parabon had expanded Snapshot to incorporate a promising new technique: forensic genealogy. Now, the same DNA sample used to produce the composite sketch of April’s killer could be uploaded to an open-source genealogy database called GEDmatch—founded in 2010 as a free service to help people look for and identify prospective relatives. The hope was that after comparing the uploaded sample to the more than one million users who had uploaded their DNA to GEDmatch, the database would spit out a list of potential relatives of the suspect.

Within months of the sample upload, Parabon’s lead genealogist, CeCe Moore, had narrowed down the DNA as belonging to two brothers: one only ever identified in court records as JPM, and his elder brother, John D. Miller, a 59-year-old third-shift Walmart worker living alone in a trailer in Grabill, Indiana, about 15 miles north of Fort Wayne. Neighbors had done their best to steer clear of Miller for years. He was strange and often angry, never marrying and keeping to himself.

Fort Wayne police were nearly certain that Miller was their killer, but they could not legally make an arrest based only on the analysis of a distant genealogist. They had to get Miller’s DNA, which they extracted from used condoms left in the trash outside his house. (Per Indiana state law, abandoned DNA does not require a warrant.) When they tested the DNA found in the condoms, it matched that of April Tinsley’s murderer.

Police went to Miller’s house on July 15, 2018, and he agreed to come to the police station to talk to detectives. At the station, they asked Miller if he knew why he was there. “April Tinsley,” he replied, and confessed to the child’s rape and murder. Miller’s criminal record had been mostly clean; he had never left the Fort Wayne area, and he had never committed other major crimes that the police knew of. “Nothing else would have drawn him to our attention,” Allen County prosecutor Karen Richards tells me. “You always hear the phrase ‘hiding in plain view.’ That’s what this guy was doing.”

April Tinsley’s murder was one of the dozens of cold cases solved with the aid of forensic genealogy in 2018. The technology had exploded into public awareness in April of that year, when Joseph DeAngelo was arrested for the crimes associated with the Golden State Killer case, which involved a dozen murders and more than 50 rapes committed in California between 1976 and 1986. Headlines proclaimed, “A Popular Genealogy Website Just Helped Solve a Serial Killer Cold Case,” and “The Future of Crime Fighting Is Family Tree Forensics.” Relatives of victims breathed sighs of relief that the perpetrator had finally been identified and caught. Law enforcement professionals and dedicated internet sleuths were also excited—not just because long-dormant criminal cases were being solved, but also because of how they were being solved. And by whom.

Forensic genealogy can best be described as a still-nascent technique of forensic science that combines DNA analysis and family-tree building. (The TV pitch would be a cross between the documentary series Forensic Files and Who Do You Think You Are?, in which celebrities explore their family trees.) Its specific alchemy results when the field of genetic genealogy—which uses DNA testing to help people discover and identify their ancestors—is applied to legal and investigative issues, like the tracking down of missing heirs, adoptive parents, and siblings, the assigning of names to the unidentified remains of soldiers, and now the cracking of cold cases.

“You always hear the phrase ‘hiding in plain view.’ That’s what this guy was doing.”

It is also a specialty of Parabon NanoLabs, which has successfully identified more than 30 suspects in cold cases since May 2018, establishing itself as the go-to service for forensic genealogy. The company was originally founded in 2008 by computational scientist Steven Armentrout and chemist Michael Norton to develop products for analyzing the tiniest amounts of DNA and applying them in cancer research, developing new vaccines, and creating novel synthetic drugs. The Department of Defense had funded the initial research that became Snapshot; law enforcement requests came later, with the Fort Wayne police becoming one of Parabon’s earliest clients in 2015.

When I studied forensic science at John Jay College of Criminal Justice in the early 2000s, I toiled—with mixed results, at best—on DNA research that convinced me I was better off far away from the laboratory, writing about crimes rather than solving them. As DNA testing became more sophisticated, gleaning results from microscopic samples that were once thought untestable, I got excited at the possible ramifications with respect to cold cases both famous (Zodiac Killer, anyone?) and unknown.

What also stands out about the rapid rise of forensic genealogy is that the work was, for years, the domain of amateur genealogists. These genealogists toil in archives and databases first for themselves, then for others who enlist their help to solve mysteries and unknowns in their family trees.

Only a handful of these genealogists, however, have the requisite expertise and background needed to resolve cold cases like those that Parabon takes on. Colleen Fitzpatrick, a 63-year-old physicist who coined the term “forensic genealogy” in the mid-2000s, is one of these experts. Fitzpatrick concentrates almost exclusively on unidentified remains, and she most recently cofounded the DNA Doe Project with amateur genealogist Margaret Press. So, too, is Barbara Rae-Venter, 70, the California-based genetic genealogist who worked with the FBI and with various divisions of state law enforcement to pinpoint the identity of the Golden State Killer. (Neither genealogist has worked with Parabon.)

Both Fitzpatrick’s and Rae-Venter’s expertise in forensic genealogy is entirely self-taught, independent of formal schooling or training, retirement hobbies turned into new careers. It’s a trajectory that CeCe Moore, Parabon’s lead genealogist, knows well. Though she became interested in genetics as a high school student, none of the universities she applied to offered an undergraduate genetics major, so she decided to study voice and later took up acting, performing leading roles in local musical theater productions like West Side Story, as well as in numerous TV commercials. But in 2008, as she was about to crest 40—Hollywood’s dreaded inflection point, where parts for women begin to disappear—Moore began to indulge her earlier enthusiasm for population genetics by tracing her own family history.

As it turned out, Moore wasn’t just interested in genealogy: she was very, very good at it. After tracing her own family history back to the 1400s, she decided she wanted to help other people solve their own ancestral mysteries. Though genetic-testing companies such as 23andMe, Ancestry.com, and FamilyTreeDNA had been around since the early 2000s, more complicated efforts—such as identifying unknown parentage, for example—required time-intensive approaches.

In 2010, Moore started a blog about genetic genealogy called Your Genetic Genealogist, which attracted the attention of other genealogists, academics, and law enforcement types for demonstrating how genealogy research could be combined with DNA testing. She soon began consulting for television programs such as PBS’s Finding Your Roots, hosted by historian Henry Louis Gates Jr. She taught classes in genetic genealogy all over the country. In 2015, Moore created DNA Detectives, a Facebook group devoted to helping people find unknown biological relatives using their own DNA. The group ended up attracting 100,000-plus members, reflecting Moore’s stature and the way in which other genealogists regard her as a leader in her field.

Ellen McRae Greytak doesn’t recall the exact moment when she realized that Moore’s expertise might prove useful for Parabon NanoLabs. Dr. Greytak, who has a PhD from Harvard University in evolutionary biology, had joined Parabon as director of bioinformatics in 2012 without giving crime-solving much thought. She had spent her academic life looking into the genetic sequences of centenarians to figure out why they, and not just anyone, lived such long lives. After finishing her postdoctoral fellowship at the Institute for Genome Sciences, Greytak was hired by Parabon to apply that work to new drug development and other research.

Greytak devoted her energy to Snapshot, which was already a couple of years into its existence when she joined Parabon, and found herself drawn to its uses for forensic science. What sometimes frustrated her about Snapshot was that its predictions couldn’t be made with close to 99 percent confidence—it was more like 80 percent, or even less. That’s because Snapshot had been designed to be used with perfect samples with lots of available DNA, while crime-scene samples are often degraded or unpredictable. But her company’s law enforcement clients kept returning with the same refrain: what can you tell us with 95 percent confidence?

The only thing the existing technology could tell law enforcement with any certainty was how to rule people out: We can tell you he doesn’t have brown eyes. But Parabon had designed its Snapshot techniques to rule people in, to identify one specific person; Greytak realized that Parabon’s technology could be used alongside existing methods to produce more specific results. “We’ve got a suspect list. We have no description. We have no way to prioritize among those people,” says Greytak. “And now you’re telling us that we can take all those brown-eyed people and put them at the bottom? That is extremely important.”

What stands out about the rapid rise of forensic genealogy is that the work was, for years, the domain of amateur genealogists—toiling in archives and databases first for themselves, then for others who enlisted their help.

Once there is a genetic profile to analyze, Snapshot’s next step is to show how that genetic profile is translated into observable characteristics, like eye and hair color. “We had to understand how people are related,” Greytak tells me, “to be able to say, ‘Okay, this isn’t our guy, but maybe his second cousin is.’”

That particular thought experiment led Parabon to expand Snapshot to include kinship analysis, which helps the researchers better predict that an individual (or suspect) is, in fact, related to the person submitting the original DNA sample. New challenges emerged for the company, especially when Parabon got a contract in 2016 from the US Army Research Office to work on unidentified remains of American soldiers; many of the DNA samples of the soldiers’ remains had degraded to the point where the noise far outweighed the signal.

Greytak needed better data to make Snapshot that much more useful a service. She needed known examples of families where the relationships were established to the level of third or fourth or even fifth cousin, where the amounts of shared DNA dropped off, but never to zero. “Where do we find large families where people know their relationships to one another? And ideally, some of them have already DNA tested so we don’t have to pay lab fees?” she says. “That’s when we thought of the genealogy community.”

Greytak started researching genealogy conferences and reading up on the names and types of experts who might attend them. She scoured message boards and blogs for places to post about Parabon’s need for multiple generations of family-history data (and the DNA sequences necessary) for the new project she had in mind. She posted requests for volunteers, some of whom might already have submitted saliva samples to private databases and received DNA results. She also looked into public databases, like GEDmatch, which allows people to find possible matches to relatives as far back as the third-cousin level.

One person immediately posted about Parabon’s project to her Facebook page: CeCe Moore. As Moore tells me, “Clearly, I need to contact these people, because I just keep hearing this Parabon name. I knew about their Snapshot technology because I’d been teaching in my classes in the future of genetic genealogy that we’d be able to use DNA to reconstruct ancestral genomes and then predict what our ancestors looked like. Here they were doing that with crime-scene DNA … something is up here.” The general response, not just from Moore but from the entire genealogy community, was enthusiastic. It made Greytak wonder: how else might Parabon and genealogists like Moore work together? CeCe Moore needed Parabon NanoLabs in order to do her work more professionally, and to solve cases from those clamoring for her help. But Parabon needed Moore just as much, if not more, to break those cases wide open.

Moore began her discussions with Parabon in earnest in the following year, detailing her interest in identifying John and Jane Does. Greytak, in turn, told her about Parabon’s existing work with the US Army and with law enforcement. In late 2017, the two women joined forces to create a separate pilot project, in which crime-scene DNA could be uploaded to GEDmatch in the hopes of identifying John and Jane Does. Identifying perpetrators, however, presented fraught ethical issues that Moore wasn’t certain how to grapple with.

“I thought it was for the greater good, and I certainly wanted to stop these criminals and save lives if possible,” Moore says. “I knew I could do it with my skills and experience, but I didn’t want to betray my community and individuals that trusted me. The people [using] GEDmatch didn’t know their DNA could be used for these purposes.”

Moore felt—as did many other genetic genealogists—that identifying potential perpetrators using GEDmatch was a no-go zone. Users had signed up for the service in order to trace their family history, not to serve as conduits for solving crimes. The ethics of using the DNA for an entirely different purpose were murky at best, queasy-making at worst. “We have to be very sure we’re not misidentifying suspects … because that’s even more devastating for a family,” she says. “Having that level of accuracy, the experience in dealing with sensitive situations, that's the most important thing.”

Then, in April 2018, came the announcement of the arrest of a suspect in the Golden State Killer case. Only weeks before, the unknown killer’s DNA sample had been uploaded to GEDmatch by detectives, and the identification of third and fourth cousins pointed law enforcement in the direction of Joseph DeAngelo. Days after the announcement, and the resulting media coverage, GEDmatch changed its terms of service to make it plain to users that their uploaded DNA could, in fact, be accessed and used by law enforcement as part of investigations into homicides and sexual assaults, and that they were free to opt out of this access if they wished.

Ethical issues assuaged, Moore officially joined Parabon in 2018 and set out to create and lead their genealogy division. The company sent out a press release about Moore’s hire. Right away, requests from dozens of law enforcement agencies across the country rolled in. Moore assembled a small team of genetic genealogists and got to work on trying to solve cold cases—first by building family trees back in time, generation by generation, and then creating lists of possible descendants in hopes of pinpointing the identity of a possible killer or serial rapist. Sometimes the work took weeks. Sometimes it took a single weekend. But Moore was stunned by how well Parabon’s genetic genealogy program helped investigators make identifications.

In a statement issued in January 2019, which announced that Parabon had solved 32 cold cases the previous year, Moore said: “We have come to expect a successful identification nearly every week.” Two weeks later, Moore and Parabon’s work led to the solving of a four-decades-old cold case, the 1979 murder of 20-year-old Anna Hlavka in Portland, Oregon. Using DNA evidence, Parabon was able to identify Hlavka’s killer as Jerry McFadden, executed in 1999 for the 1986 rape and murder of a teenage girl in Texas.

Because all of the suspects in cold cases fully resolved by Parabon have either pled guilty or were deceased by the time they were identified, forensic genealogy has yet to be tested in a courtroom. (This past December, John D. Miller pleaded guilty to April Tinsley’s murder and was sentenced to 80 years in prison.) Once it is, the question becomes whether or not the technology will be deemed robust enough to withstand the Daubert standard, a key test for scientific techniques named for a 1993 Supreme Court ruling, which is supposed to determine whether a branch of forensic science is “generally accepted” in the scientific community. (DNA testing met the Daubert standard right away, in large part because it was a new-enough technique to have its merits and flaws scrupulously examined by prosecutors and defense attorneys alike.)

When forensic genealogy is subjected to interrogation in a courtroom, perhaps as early as this year, how will Parabon handle it when defense attorneys put its scientific approach under the microscope, perhaps aggressively? The answer, Greytak tells me, is to to focus on the limitations of the technology, to stress that genetic genealogy is, for all intents and purposes, a presumptive test, not a confirmatory one. “Even though, sometimes, our information results in a name that we’re giving to the police, it’s still [just] an investigative lead,” she says. “That’s the way we always couch it. We are recommending that you look at this person, but we can’t say so with 100 percent certainty.”

Once law enforcement receives that lead—one that Parabon cofounder Steven Armentrout tells me is “no different from an anonymous tip or from a composite sketch artist”—they still have to do the work to investigate and obtain DNA from that person of interest, as was the case with April Tinsley’s killer. “Our information helped [law enforcement] locate that person, but it’s the DNA match that’s actually used in the prosecution,” says Greytak. “We do recognize that, since our information was helping them to get there, that could be part of the court case. But the end result is to say that the DNA matched.”

Although Parabon NanoLabs is the only private company in the world to employ forensic genealogists, that’s about to change. This past February, forensic-science-services company Bode Technology, based in Lorton, Virginia, announced the formation of its own proprietary forensic-genealogy service, and more private companies may follow suit, meaning that the the field of forensic genealogy may look completely different a year or even a few months from now. Other genealogy databases may join GEDmatch as a wholly public offering, where anyone can upload their DNA and discover the answers to their ancestral mysteries, one generation at a time, by building family trees, up and down the centuries. Parabon and other labs may also begin to prioritize cases still in the active stages of investigation, rather than those that are ice-cold, which would make genetic genealogy more of a first line of defense, rather than a last resort.

As with the initial use of DNA testing in a forensic science capacity three decades ago, more practical concerns will appear as forensic genealogy becomes better understood and more widely accepted. The technology as done by Parabon is expensive, each analysis costing upwards of $5,000. Multiply that by the number of open cases, cold or hot, and the math doesn’t work in investigators’ favor.

Parabon’s researchers focus on the limitations of the technology, to stress that genetic genealogy is a presumptive test, not a confirmatory one.

Some agencies, like the Florida Department of Law Enforcement, are working around the expense by creating their own forensic genealogy units. But most law enforcement agencies simply can’t afford to use genetic genealogy on every cold case, let alone newer ones. “Maybe someday this technology can be done by state forensic labs, but right now we’re still struggling to get analysts to do the basic DNA testing,” Karen Richards, the Allen County prosecutor who oversaw the April Tinsley case, reminds me. “Results take months and months to come in. So until we get a handle on more adequate and timely basic testing, I can’t imagine trying to add another layer of complexity.”

In the meantime, there will only be more questions about the validity of forensic genealogists working with law enforcement, about ethics and privacy concerns, and about who is qualified to make those calls. In January 2019, BuzzFeed broke the news that FamilyTreeDNA, one of the oldest private at-home genetic-testing companies—and the one that first attracted CeCe Moore’s attention as a budding amateur genealogist—has been allowing FBI agents to search its genealogy database in an effort to solve violent crimes, a shift that is currently both legal and welcomed by the company.

FamilyTree, established in 1999, had changed its terms of service in December 2018 to assert that law enforcement could make use of its data for homicide and rape cases, but for the more than a million people whose DNA samples became part of FamilyTree’s database in its 18 years of existence, finding out the FBI had access to their data without their permission was not what they had signed up for. As Alan Butler, senior counsel for the Electronic Privacy Information Center, a nonprofit research center focusing on privacy issues, told the New York Times in February: “The company needs to either roll back the change or else delete all stored DNA data it has collected from individuals under the previous agreement.” Already, one bill to limit the use of forensic genealogy in cold cases, or ban it outright, is being considered in Maryland. More states may follow suit if public outcry grows.

Forensic genealogy may still be in uncharted, unregulated territory, but as applied to crime-solving, it looks to be as paradigm-changing as DNA testing was to forensic science—like a light switch flipped on what was previously thought of as of as permanent darkness. But the technique is new enough that the serious concerns already lodged about ethics and privacy should not be ignored. We want crimes to be solved, but at what cost? Is the excitement of resolving a long-unsolved murder worth the price of ever-increasing police surveillance, or falling into entrenched racial biases, or prioritizing arrests over community, prison over rehabilitation?

Thinking about these questions tempers my own excitement about genealogy’s present and future role in law enforcement. The quest to understand our ancestral roots is filled with the landmines of unintended consequences. Because, as genealogists well know, DNA is a shared experience. One DNA upload by one person curious about their ancestry leads to the arrest of a killer in a decades-old case. That DNA upload comes from a specific individual, yes, but each generational remove creates a community. It’s not just your DNA, or mine. It’s all of ours, flying out of that Pandora’s box, and we have no real idea where it may lead next.