Law Enforcement And Genealogy Databases
Yesterday we talked about Microarrays, arrays appear to retain a strong market in genealogy (ancestry.com/23andMe). So, I thought today it might be interesting to take a quick look at how that data can be used for… other applications.
GEDMatch is a service (acquired by Qiagen) which allows you to upload your (microarray) data from 23andMe and other services and use this information to find lost relatives. This information can then also be used by law enforcement, with (or without) user agreement. Most famously this was used to help identify the Golden State Serial Killer, here user agreements appear to have be retroactively updated to allow use by law enforcement in this specific case.
Finding a suspect with GEDMatch isn’t as easy as uploading a crime scene sample and getting an exact match, but it can reconstruct a family tree, or eliminate suspects. In the Golden State Serial Killer case they tested at least one other suspect before deciding they’d identified a match. As well as eliminating suspects, I assume these samples can help more accurately locate the target individual within the family tree.
In the Golden State Serial Killer case, once they’d figured out who they were after, they obtained discarded tissues and door handle swabs to confirm the match before making an arrest…
This is fine as it goes, but there’s now an ongoing lawsuit between a number of individuals and GEDMatch/Verogen….
According to one lawsuit1 GEDMatch provided law enforcement access to data even when users “had expressly instructed Verogen not to disclose their genetic information to law enforcement entities.”
GEDMatch has a service for law enforcement called GEDMatch PRO (launched after the Golden State Serial Killer case). For $700 law enforcement entities can provide a sample from a crime scene and this will be matched against the GEDMatch database.
Users should have to specifically opt in to this service. But a press release on the Verogen website states that:
“We recently learned that a small number of forensic genetic genealogy practitioners had circumvented GEDmatch settings in violation of our Terms of Use, enabling them to access some profiles of GEDmatch users who had not opted in to law enforcement investigations for violent crime and homicides.”
This stuff was all new to me, but the intercept wrote about it ages ago apparently. The exact nature of the loophole isn’t described, but the intercept article implies it was somewhat intentional.
The whole nature of these databases is to let users upload their genetic profile and search for relatives. So it seems perfectly possible that law enforcement could just use that route if they wanted to. And they appear to have used in this in some instances.
Personally I suspect keeping genetic information out of the hands of law enforcement is going to be impossible in the medium to long term. The only practical solution would be blanket ban on law enforcement use of genetic databases… and I can’t see that happening.
Let’s run the numbers and try and figure out how difficult it would be to make a near 100% effective law enforcement database.
The GEDMatch match for the Golden State Serial Killer was “a probable 4th cousin”2. Elsewhere it’s been estimated that databases of 1.3M individuals could narrow a suspect down to”fewer than 20 people” in about “60% of white Americans from a DNA sample”.
From that I’d roughly guess you need about 1% coverage of a population to identify nearly every suspect3.
The thing is, this doesn’t need to come from the current population… if we assume we can spread this over the last 4 generations then we only need 0.25% of the population to be in databases law enforcement can access.
Given ~1% of the US population is in prison, and that law enforcement DNA-profile databases already exist. Building a database of this scale doesn’t seem intractable.
You just need to wait 100 years…
Whether you think this is a powerful tool to identify criminals, or an altogether too powerful tool with potential for misuse is an open question.
But without legislation it seems almost inevitable that these database will exist, and be very effective at locating individuals.
In you’re interested in the full filing in the GEDMatch case, it’s linked below: