Study: DNA web­sites cast broad net for iden­ti­fy­ing peo­ple


NEW YORK About 60 per cent of the U.S. pop­u­la­tion with Euro­pean her­itage may be iden­ti­fi­able from their DNA by search­ing con­sumer web­sites, even if they’ve never made their own ge­netic in­for­ma­tion avail­able, a study es­ti­mates.

And that num­ber will grow as more and more peo­ple up­load their DNA pro­files to web­sites that use ge­netic anal­y­sis to find rel­a­tives, said the au­thors of the study re­leased Thurs­day by the jour­nal Science.

The use of such data­bases for crim­i­nal in­ves­ti­ga­tions made head­lines in April, when au­thor­i­ties an­nounced they’d used a ge­netic ge­neal­ogy web­site to con­nect some crime-scene DNA to a man they then ac­cused of be­ing the so-called Golden State Killer, a se­rial rapist and mur­derer.

In gen­eral, such searches be­gin on a site by find­ing a rel­a­tive linked to a DNA sam­ple. Then sleuths can use other in­for­ma­tion like pub­lished fam­ily trees, pub­lic records and lists of sur­vivors in obit­u­ar­ies, plus what­ever they know about the per­son whose DNA be­gan the process. They can build their own spec­u­la­tive fam­ily trees. Even­tu­ally, that can point to some­one whose DNA is then found to match the orig­i­nal sam­ple.

With DNA data­bases “you need just a minute frac­tion of the pop­u­la­tion to re­ally iden­tify many more peo­ple,” said Yaniv Er­lich of Columbia Univer­sity, an au­thor of the study.

Each per­son in a DNA data­base acts “as a bea­con that il­lu­mi­nates hun­dreds of dis­tant rel­a­tives,” said Er­lich, who is also chief sci­en­tific of­fi­cer of the MyHer­itage web­site.

His pa­per fo­cused on Amer­i­cans of Euro­pean de­scent be­cause such peo­ple are over-rep­re­sented in DNA data­bases, which makes it eas­ier to find rel­a­tives.

The re­searchers started with the 1.28 mil­lion par­tic­i­pants on the MyHer­itage site at the time they did the work. Most had a north­ern Euro­pean ge­netic back­ground. For each, they looked for rel­a­tives more dis­tant than first cousins else­where in the data­base.

About 60 per cent of the time, they found some­one whose ge­netic sim­i­lar­ity was at least equal to that of a third cousin, sim­i­lar to the de­gree of re­lat­ed­ness that led to the Golden State Killer sus­pect. Third cousins share great-great-grand­par­ents.

With some ba­sic as­sump­tions about what kind of data would be avail­able for a crim­i­nal sus­pect, the re­searchers cal­cu­lated they could pare down the pos­si­ble iden­tity of the ini­tial per­son to just 16 or 17 peo­ple. That’s lim­ited enough that po­lice could zero in with fur­ther in­ves­ti­ga­tion, Er­lich said.

Er­lich and his co-au­thors sug­gested that such searches could cast a broader net in the near fu­ture. A data­base with DNA pro­files of just 2 per cent of a pop­u­la­tion is enough to match nearly every­body with some­body who’s as closely re­lated as a third cousin, re­searchers said. From that, they cal­cu­lated that the ge­netic pro­files of about 3 mil­lion Amer­i­cans of Euro­pean de­scent could de­liver the equiv­a­lent of a third cousin for more than 90 per cent of that eth­nic group­ing.

Web­sites are get­ting very close to that, said Er­lich, not­ing that MyHer­itage now has more than 1.75 mil­lion par­tic­i­pants. He said the web­site does not al­low foren­sic searches.

Newspapers in English

Newspapers from Canada

© PressReader. All rights reserved.