[INTRO ♪].

Back in the ‘70s and ‘80s, a man known as the ‘Golden State killer’ terrorized California, committing a string of more than 100 burglaries, 45 rapes, and a dozen murders. Definitely a bad dude.

Even though the police had DNA evidence from crime scenes, it never matched any DNA records on file, so the case went cold ... Until last week. By this point, you’ve probably heard about this man’s capture.

But you might not realize that the science used to crack the case is totally different from standard DNA forensics. So by understanding how DNA analysis is typically done for crime scenes, we can dive into why the Golden State killer case is so special. Usually, forensic scientists process DNA with a method called short tandem repeat, or STR, analysis.

The basic idea is that little stretches of short repeated sequences, like TAGA, are scattered throughout your genome, in specific places on specific chromosomes. Because the number of repeats at each location tends to vary from person to person, counting them can be useful. Like, they can help match the semen in a rape kit to a suspect.

For instance, you might have 6 TAGA repeats on a chromosome that you got from one parent, and 10 repeats from your other parent, while I might have 12 and 15. Now, if the number of repeats at one spot in someone’s DNA happens to match evidence from a crime scene, that doesn’t necessarily mean much. But if you look at a bunch of repeats and they all match, you might be onto something.

In the US, forensic scientists have traditionally tested for repeats at 13 different spots in the genome, although recently they upped it to 20. That’s what’s happening when you hear about hits in CODIS, the FBI’s database of DNA records. A match is all about probability, and there are a lot of factors to consider, like how common certain STRs are in a given population.

But if your suspect’s DNA repeats in the same 13 ways as a sample from the crime scene, the odds that the suspect isn’t the source are typically about one in a billion. In other words, combined with other evidence, you can be pretty confident that you’ve got the right person. Of course, DNA analysis isn’t perfect.

Contamination is a concern, and there’s room for doubt in cases where there’s a small amount of DNA, or the DNA is degraded. These days, standard STR analysis uses a process called PCR to amplify the DNA sections the police are interested in. But that can lead to errors, and they might not get solid data for all 20 repeat locations.

Another challenge is that many samples include DNA from multiple people, like both a victim and the criminal, making the analysis more complicated. And any positive match still needs to be interpreted. After all, an innocent person might have scraped their finger and left a few drops of blood behind at a future crime scene.

So that’s how standard forensic DNA testing works. And if it’s all done right, it can be really persuasive, as you probably know from shows like CSI. But with the Golden State killer, the cold case heated up because of a different approach to DNA— one that’s a lot closer to spitting in a tube to find out your ancestry.

Detectives had some DNA evidence that was collected from a double murder in 1980 and frozen, so it was especially well-preserved. We don’t know the specifics of what they did next, but they eventually uploaded data from that sample onto an open-source genealogy website called GEDmatch. While GEDmatch isn’t a power player in the commercial DNA industry like 23andMe or Ancestry.com, it runs on the same kind of information.

So someone might use one of those services to get raw genome data, then submit it to GEDmatch to help find long-lost relatives and piece together their family trees. The FBI created the DNA profile of the Golden State killer in their own labs, from that well-preserved sample. And they probably generated the same kind of data, by looking for single nucleotide polymorphisms, or SNPs.

These are individual As, Ts, Gs, and Cs throughout the genome that we know vary between people and can be passed down from parents to kids—and because of that, they’re useful biological markers. In a way, SNP analysis is similar to STR analysis, except it’s cheap and easy to test for tens of thousands of these puppies at once. So that’s what’s going on when you mail in your DNA.

Most companies aren’t directly sequencing your entire genome. Instead, they’re using technology that checks your DNA for a whole bunch of SNPs. Most importantly for detectives, because there are so many SNPs, and they change less over time from mutation,.

SNP testing is much better than STR analysis when it comes to identifying far-flung relatives. So, once the detectives on the Golden State killer case uploaded the genetic profile of their suspect, they looked for similarities among GEDmatch’s 900,000 other profiles. Their killer hadn’t uploaded his own data, but some of his distant relatives— like, third and fourth cousins—had.

These are people who would share about 2% of their DNA with him at most. But it was enough of a lead to build potential family trees—some 25 in all. Within 4 months of getting some initial hits on the genealogy site, police officers had painstakingly narrowed their focus from thousands of relatives to one man.

He was the right age, and had lived in California during the crime spree. To confirm that he was the Golden State killer, they needed some of his DNA to test for a match. So they put him under surveillance and grabbed his trash.

They used something with some of his cells on it: maybe a straw, soda can, or used Kleenex—we don’t know what exactly. And then they did a DNA test, probably STR analysis, and got a match. Just to be sure, they checked again.

Another garbage item, another test—and another match. After 44 years, they had identified the Golden State killer. He was a former cop named Joseph James DeAngelo, now age 72, living in a suburb of Sacramento, California.

Now, this case isn’t the only time DNA has been used like this, but it’s one of the highest profile cases. And it’s likely to have people talking for a while, especially about data privacy issues. We won’t get into that here, but it’s definitely something people are thinking more about, as investigators realize how many ways they might be able to use DNA.

Even if it involves way more work than TV crime dramas ever show. Thanks for watching this episode of SciShow News! If you want to learn more about forensics, you can check out our episode that dives into a lot more crime scene science.

And if you just want to keep thinking about the world more complexly with us, you can go to youtube/scishow and subscribe. [OUTRO ♪].