Campus & Community

Catching criminals through their relatives’ DNA

5 min read

Finding genetic needles in database haystacks

Deborah Sykes was on her way to work at the Winston-Salem Sentinel newspaper in North Carolina on the morning of Aug. 10, 1984. She parked her car and began walking the few blocks to her Sentinel office. She never made it.

The 25-year-old woman was raped, sodomized, and stabbed to death.

About a month later, police arrested 19-year-old Darryl Hunt. Although he maintained his innocence, Hunt was convicted of the crime and sentenced to life in prison.

But in 1990, Hunt caught a break. By then, reliable tests were available that could compare DNA in semen found on Sykes with that of her attacker. Testing in 1990, and again in 1994, showed that Hunt could not possibly be the source of the semen. Both times, law enforcement officials ignored the evidence.

Frederick Bieber and his colleagues used a computer experiment to find matches to DNA found at crime scenes, even though police had no exact matches in their criminal databases. (Staff photo Rose Lincoln/Harvard News Office)

Further investigation, however, found a close, but not perfect, match to the genes of another man, whose DNA was in a database of criminals kept by the state. More police work led to the questioning of this man’s brother, who had once attacked a woman not far from the place where Deborah Sykes was killed. Confronted with DNA evidence, the man, Willard Brown, confessed to Sykes’ murder on Dec. 22, 2003. In February of the next year, Darryl Hunt finally walked free after spending half of his life in prison for a crime he didn’t commit.

This case is cited by Harvard University researchers to demonstrate the potential of using the DNA of relatives to catch guilty kin. “Close relatives have particularly similar DNA profiles due to shared ancestry,” notes David Lazer, an associate professor of public policy at the Kennedy School of Government. “As the Sykes case illustrates, this could be exploited in criminal investigations.”

99 percent success

Lazer, along with colleagues Frederick Bieber, a pathologist at Harvard Medical School, and Charles Brenner, a mathematician at the University of California, Berkeley, designed computer experiments to demonstrate their point. The team started with a DNA sample from a crime scene that failed to be an exact match to any profile in a large, up-to-date database. Their computer program compared this sample to the DNA profiles of 50,000 offenders in what they call “a typical-sized state database.” Offenders with the closest matches would, in an actual forensic investigation, be singled out as possible relatives of the suspect.

From this simulation, they concluded that a parent or child of the suspect would be identified as the first candidate more than half (62 percent) of the time. Looking at the top 100 candidates, they raised their success rate for finding the best candidate to 99 percent. Further analysis showed they could also identify brothers and sisters of the suspect whose DNA they had, albeit at a slightly lower success rate.

Their report, published in the online version of the May 12 issue of the journal Science, included discussion of another case where DNA of a relative led to solving a brutal murder. Lynette White, a 16-year-old girl, was found with her throat slit and more than 50 stab wounds on Valentine’s Day 1988, in Cardiff, Wales. The crime scene yielded plenty of blood, including a sample that contained a rare form of a gene, which also appeared in the DNA of a 14-year-old boy whose profile was on file at the United Kingdom National Database.

Fourteen years after the actual crime, committed before highly accurate DNA profiles were available, police concluded that the murderer was a male relative of the boy. They asked several relatives, including an uncle named Jeffrey Gafoor, for a DNA sample. Gafoor, believing the police were trying to match a semen sample, agreed and told police that he had had sex with Lynette White around the time of the murder. It was, he thought, a clever way to account for his semen at the murder scene. But the evidence came from a blood sample and it matched his DNA profile perfectly. After a long and frustrating search for White’s killer, which included the imprisonment of three innocent men, the mystery was solved in 2003.

Ifs and buts

Needless to say, kinship DNA mapping is not going to solve every crime. Its success depends on the presence of the DNA of a close relative of the guilty person being in a database of criminal offenders. Addressing that probability, the researchers mention that a U.S. Department of Justice survey found that 46 percent of jail inmates said they had at least one close relative who had been incarcerated.

Banking DNA from criminals and arrestees has been challenged as violating Constitutional protection again unreasonable search and seizure. As Lazer points out, “such challenges have not prevailed, as the courts have ruled that the interests of public safety outweigh individual privacy interests.” However, the last legal word has not been spoken as attempts expand to cover those convicted of minor crimes and misdemeanors.

The Harvard researchers recognize that identification of kin by searching DNA databases of criminals could lead to widespread opposition. However, it also might lead to louder calls for a universal DNA database, which to date have been rejected. Those in favor of such expanded databases insist that intrusions of privacy could be limited by restricting familial searches to only the most serious crimes, such as those committed against Deborah Sykes and Lynette White. Database search methods could also be defined to minimize restrictions on innocent parties.

Finally, as Bieber emphasizes, “It’s crucial to note that leads derived from DNA matches do not necessarily imply guilt or lead to conviction.” Still, as in cases like the murders of Sykes and White, family DNA dragnets can significantly improve the chances of finding the genetic needle in a haystack of frustrating clues and false arrests.