Investigators trying to ID an unknown person may spend months combing through massive family trees. Credit: Cory Hall/Stanford Graduate School of Business
A research duo who has previously applied their mathematical expertise to improve forensic processes such as fingerprint, sexual assault kits and ballistics analysis has identified their next area of interest—forensic genetic genealogy.
The new method proposed by Stanford University professors Lawrence Wein and Mine Su Ertürk aims to maximize the probability of finding a descendant solution in the shortest time possible. In simulated runs using data from the DNA Doe Project (DDP), the team solved cases 10 times faster than the current average.
A typical genetic genealogy investigation begins with a target DNA sample, typically one from an unidentified body or one left by a suspect at the crime scene. From there, it’s uploaded into a DNA database that generates a list of “matches”—which is where the puzzle starts for genealogists.
Matches could be in the hundreds or thousands, usually distant cousins whose shared ancestors may have died more than a century ago. Genealogists then go about constructing a massive family tree that ties these far-flung relatives to the person who supplied the sample DNA.
But, it’s not immediately apparent which matches will provide the best path to the target. This is where a genealogists’ instincts and perhaps experience come into play—as well as the hundreds to thousands of hours spent rifling through public records.
And while this method has seemingly worked since it took off in 2018, Wein says the “decentralization” of the technique is a fault—one that can be improved through mathematics.
“You have a team of people doing this and they will each decide to take a match to investigate, and then they’ll go off on their own to try to build a family tree backward in time from each match,” he said. “They’re not thinking about the big picture holistically. Basically, we’re telling them, ‘Given where you are in the search right now, this is what you should do next.”
In their study, published in the Journal of Forensic Science, Wein and Ertürk used simulated versions of 17 DNA Doe Project cases, 8 solved and 9 unsolved. The cases Wein and Ertürk analyzed had between 200 and 5,000 matches.
While the current method looks for common ancestors between different matches, the new Proposed Method finds the most recent common ancestor between a match and the unknown target. After identifying a list of possible most recent common ancestors, the method “aggressively” fills out the family tree with their descendants, even if there’s only a slight chance that the target’s ancestor is on the list.
“We do this by describing the reconstructed family tree as a collection of probabilities that represent how likely each person on our tree is to be a correct ancestor of the target,” explains Ertürk. “Then, looking at these probabilities, you can tell which parts of the tree you should explore more.”
According to the study, after running hundreds of simulated searches, the authors conclude their method can solve a case with a 7,500-person family tree around 94% of the time. The standard method’s success rate in those cases is about 4%.
Wein and Ertürk did note their Proposed Method does not account for the genealogist him/herself, or the “tricks” they use to narrow down the options, such as focusing on family members who lived in a specific location at a specific time.
“In no way is our algorithm meant to substitute for genealogists,” said Wein. “But if they’re really stuck, it will give them some ideas that may not be obvious.”