Over time, GEDmatch has become the go-to destination for serious genetic genealogists. People have found distant family members on the site, adoptees have found their biological parents, donor-conceived kids have found their sperm donors. It’s no wonder the police came calling, too.
When the FBI created CODIS in the 1990s, the way to identify people by DNA was looking for short tandem repeats, or STRs. As their name implies, these are short sequences that can be repeated dozens or hundreds of times in the genome. The number of repeats is highly variable from person to person. STRs are not genes, though, and they reveal little about the appearance or medical conditions of a person, sidestepping some of the privacy concerns of police collecting DNA. Today, CODIS contains STR profiles of over 16 million offenders and arrestees. Each profile looks for STRs in up to 20 locations in the human genome.
In contrast, the DNA profiles on GEDmatch contains information at some 600,000 or so locations in the genome. These profiles come from customers who’ve tested with commercial companies like Ancestry and 23andMe, which identify the genetic letters that appear in those 600,000 locations by looking for SNPs, or single-nucleotide polymorphisms. SNPs are not as variable as STRs, but you can test a lot more of them. They can also be in the middle of genes; that is how 23andMe tells you if you have a certain breast-cancer gene variant. And with enough SNPs, you can use it to trace the geographic origins of ancestors and find distant relatives.
Recently, law enforcement has gotten interested in using SNPs for leads in cold cases. Forensics labs are validating tests for SNPs that can reveal general information about geographic ancestry or physical traits like eye and skin color. But this work—cutting-edge by forensics standards—only looks at a few dozen to a couple hundred SNPs compared to the 600,000 of consumer DNA tests.
Forensics has been slow to adopt recent advances in genomics—for good reason, given the high stakes of a criminal case. But it does mean that genetic genealogists who use DNA to find family members on GEDmatch are far ahead of forensics labs.
“Law enforcement continues to use outdated DNA databases, and I don’t see a movement toward using better DNA databases,” says Blaine Bettinger, a genealogist and lawyer affiliated with GEDmatch. Bettinger says he would prefer that police build their own SNP database, where searches can be appropriately regulated. Right now, there is essentially no oversight on when and how police use GEDmatch, which is, after all, a public database open to anyone. (In contrast, using CODIS to look for close family members is regulated state by state, and many do not allow it at all.)
It’s hard to argue against using a genealogy site to catch a serial killer and rapist like the Golden State Killer. But what about less serious crimes, like drug offenses, asks Bettinger. “I think that’s just overreach,” he says. “That makes me uncomfortable. We leave DNA everywhere we go. Everywhere we touch has DNA. There’s got to be a limit.” GEDmatch’s terms of service tries to limit law enforcement use to “violent crimes” defined as homicide or sexual assault, though the site as no way of verifying that.