The “Grim Sleeper” serial killer (Lonnie David Franklin Jr.) was caught in California when police made a “Familial DNA” search of their state database. Essentially, a direct match of DNA wasn’t found, so police received permission to reduce the criteria for matching in the hope of finding a close blood relative. By luck, the killer’s son had been incarcerated, and police were soon able to apprehend the father and charge him with the murders.
I was listening to experts discussing the process, and was surprised to learn how it works. The first run of the partial match revealed some 100 possible candidates. Then a second run was performed on these 100 to find shared-Y chromosomes (since they knew the killer was male).
My question is, why wouldn’t they just run the shared-Y search from the start? My guess is this is a much lengthier process than the standard search, but isn’t DNA matching just a computer database search?
If you run the very narrow search initially, you may end up with no matches at all, and you still have to run the wider search. Running the wider search gives a number of candidates who may or may not be relatives of the offender, and then requires lots of additional police legwork to narrow down to possible offenders. However, given that you have some matches, there is now a possibility of a closer Y-chromosome match, so it is worth doing that to get a closer match - if the son (or other male lineal descent relative) wasn’t in the database, then the police have to work on the initial hundred matches, and the additional DNA test is an extra cost to the investigation. As it was, they hit pay dirt.
There was a case in the UK several years ago - someone threw a brick off a bridge on to a motorway (M3). The brick smashed the windscreen of a truck and hit the driver in the chest, who managed to drive the truck onto the hard shoulder and call for help before he died. I saw the truck on my way to work. They got a trace DNA profile from the brick, found a partial match in the DNA database, and from some family research arrested and charged the perpetrator. I know that many people are uncomfortable (in the UK) with non-offenders DNA being retained in the DNA database (i.e. DNA from non-charged suspects, or supplied for exclusion purposes), but the police want to be able to cast their net as wide as possible when they do not have a direct DNA match.
I’m not at all certain but suspect the problem was this: the forensic database stores relatively few characters, with very few or none from the Y-chromosome. In the Grim Sleeper case, after the “familial search” produced candidates, the stored samples from those people were retested for Y-chromosome characters.
(The news article isn’t quite clear, so, as I say, I’m guessing.)
There’s a potential for useful matches that wouldn’t have showed up in the Y chromosome. For instance, a half-brother with a different father would have a different Y, as would an uncle or a nephew on a different male line (roughly speaking, with a different last name). Plus, of course, any women whom the police might have had samples from (they’d have many more men, of course, but there have to be at least some women in the database).
Yeah, the standard US forensic database only includes 13 markers, with just one on the Y chromosome. There are many millions of other potential markers that can be tested if you’ve got the original sample. But the standard protocol only includes those 13. That’s good enough to match an individual with pretty high confidence, but not quite good enough to pick out relationships between family members.
From the replies above, I gather a person’s full DNA sequence isn’t stored in the typical forensics database (only ~13 markers), and that once the first round was completed and some of those excluded based on traditional evidence, they needed to retest the original samples (or get all-new ones) from this reduced group to do the Y-chromosome search.
Is this correct? If so, I can see why the process is controversial (on first reading it I couldn’t understand why someone would object, but that’s a question for GD).