Skip to main content

The Flaws of Familial DNA Matching

New research warns investigators of inaccurate results, and unfairness to minorities.


It’s always been true that a family member’s guilty conscience could be a criminal’s ultimate downfall. Now, thanks to familial DNA matching, a brother or sister can be a snitch without ever saying a word. But what about that criminal’s cousin? Half-cousin? Third cousin twice removed?

Virginia’s Department of Forensic Science confirmed on Monday that it had facilitated the arrest of a suspect through this very new and very controversial forensic tool. “Familial” DNA searches are sometimes used to help identify suspects whose DNA isn’t on record in agency databases through a near-match between crime-scene DNA and that of a suspect’s close relative whose DNA is on record. This would never be the only evidence used in a case, but it can help to focus an investigation and to avoid dead-ends when exact matches aren’t possible.

The U.K.’s first arrest based on a familial match was announced in 2004, following a random act of violence involving a brick thrown from a bridge onto a car driving below. The accused, who had no prior criminal record, was nevertheless identified because DNA evidence on the brick matched the DNA of "a close relative" whose DNA was on file. When confronted by the police, the alleged brick-thrower provided another DNA sample that proved to be an exact match to the crime DNA, and then apparently confessed.

At least 200 cases in the U.K. have involved the familial matching method, but in the U.S. it has not yet been widely used. Virginia Governor Robert McDonnell only authorized it in 2011, according to the Richmond Times-Dispatch. Texas, California, and Colorado are the only other states in the U.S. that currently acknowledge using it. In Virginia, at least, this isn’t a tool that investigators plan on pulling out of the box for any petty thief, not least because it is so expensive to use:

Under Department of Forensic Science policy, familial searching is performed only to help solve violent crimes in which public safety remains at risk, other investigative leads have been exhausted, and a chief law enforcement officer makes the request.

Most famously, familial DNA matching was credited with catching California’s “Grim Sleeper” in 2010, after the serial-killing suspect had evaded cops since 1985. Another notorious murder case, involving the 1960s-era "Boston Strangler," was finally put to bed just last month when a familial DNA match confirmed that Albert DeSalvo was the killer. DeSalvo had been suspected, but never tried for the murders, though he was convicted of other crimes and was later killed in prison. That DeSalvo died in 1973 did not prevent investigators from matching the DNA from old crime scene evidence to DeSalvo’s brother's discarded water bottle in 2013.

Despite the incredible promise of this new method, however, the law enforcement agency and the prosecutor involved in the Virginia case have been reluctant to announce who the suspect is, or even what type of crime was involved. They have only said that there were several cases that they were now able to link together with new DNA evidence. A Richmond criminal-defense lawyer and DNA expert, Betty Layne DesPortes, told the Times-Dispatch that this lack of disclosure could mean that the prosecution is afraid that the defense will focus its argument on the controversial nature of familial DNA matching.

Maybe that’s wise, too, because familial matching happens to be under the microscope right now. A new study published last week in the journal PLOS ONE raised doubts about the method’s accuracy and efficacy. According to researchers at the University of California-Berkeley and New York University, led by Rori Rohlfs, familial DNA searching will often indicate that two people are close relatives when they are in fact distant relatives. For instances, in an experiment that tested the process of familial DNA matching in the California DNA database (using simulated genetic profiles based on publicly available data), the researchers found that cousins could be misidentified as siblings.

This type of mistake could cost investigators a lot of time and money, as well as intrude on innocent people’s lives and privacy. Investigators may be interrogating immediate family members who have nothing to do with the crime. Or, the familial search could implicate very distant relatives—so distant that their relationship with the suspects will not be helpful to investigators at all.

"Both of these consequences exacerbate the numerous ethical problems presented by familial searching," the researchers write. "The greater the number of persons involved, and the less likely that one of them is in fact the perpetrator, the more such investigations may begin to feel like a fishing expedition rather than a reasonable search."

The study brings up another important consideration about familial searches: because the accuracy of the matching depends on the number and types of samples that already exist in the database, the accuracy of the matching will vary with ethnicity.

For instance, it’s a fact that African Americans are disproportionately represented in the criminal justice system as well as in crime-fighting databases. In this experiment, "while the overall rate of false identification of unrelated individuals remains low," the rate of false positives of African Americans was "much higher, roughly two orders of magnitude higher" than other groups. (By comparison, there were relatively few Native American DNA samples in their sample group; the familial matching of Native American samples produced no false positives.)

If African Americans suffer disproportionately from false-positive matches, and also from very-distant-relative matches, it follows that they will suffer disproportionately from intrusions of privacy and police interrogations:

Limiting investigations to the immediate family members of known offenders at least minimizes the intrusion on innocent relatives within those racial groups. But if more distant relations are included, the web of potential ‘genetic suspects’ becomes still broader, and may effectively encompass entire communities. It takes only one member of a large and varied family tree to render every father, brother, half-brother, cousin, half-cousin, uncle, nephew and so on vulnerable to scrutiny and surreptitious sampling by law enforcement officers.

So add these to the litany of problems with DNA collection and identification that law enforcement agencies and courts already have to contend with, including but not limited to contamination, lab errors, and privacy concerns. DNA evidence can do a lot of good: it has allowed hundreds of wrongly-accused prisoners to finally go free, and it has correctly identified countless dangerous criminals as well. But it is a very new science. And every new branch of science brings new wonders (which will likely be challenged in turn): another study out this month from the University of Kent promises a way to identify which brand of lipstick a person was wearing at a crime scene, without even removing the lipstick-stained glass or cigarette butt from its evidence-collection bag.