DNA evidence fakable.

[Darth Wagtaros]

This is bad.  I am concerned.  What are the moral ramifications of re-enacting the Lizzy Borden case with faked DNA evidence?
From Ars Technica:

If there's one application of modern genetics that the public has not only accepted but embraced, it's the use of DNA testing in criminal investigations. Courts have accepted DNA evidence as definitive, and it's difficult to imagine a movie or TV show that focuses on law enforcement but declines to use DNA testing as a plot device. The reason is simple: given a valid DNA sample, the tests can match it to its source with probabilities that exclude the rest of our planet's population. Those probabilities still hold, but some researchers have now looked into whether it's possible to fake a valid DNA sample, and they have come up with a disturbing answer: just about any molecular biology lab has the tools to do so.


DNA evidence is appealing largely because it's rigorously scientific: we have empirical data about the prevalence of different genetic variations in the population, and it's easy to calculate the probability of any individual carrying a specific combination of those variants. Look at enough of them, and you can lower that probability to the point where it's less than one in several billion, meaning that DNA's owner is likely to be the only person on earth with that precise combination.


In contrast, as recent reports have indicated, there's not a lot of science to the rest of forensic science. As the authors of the new paper point out, in contrast to the science behind DNA, "other types of forensic evidence, such as ballistics, blood-spatter analysis, and fiber analysis... rely on expert judgment and have limited connection to established science. [DNA] is even considered to be more reliable than eyewitness evidence, which is known to suffer from a relatively high rate of errors."
Which is what makes their report disturbing. The forensic processing pipeline includes a variety of methods to identify likely sources of DNA, ranging from identifying obvious points of contact like the grip of a gun, to locating sources of biological material, such as blood and saliva spatters. DNA is then isolated from these sources. But there's no way to determine if the DNA that actually wound up in or on the sample is actually the same DNA it started with.


As the researchers demonstrate, it's possible to exploit this loophole with a vengeance. Purified DNA can be smeared all over the surface of your choice, such as a gun grip. It's also possible to eliminate the original DNA from blood and saliva samples using a standard piece of lab equipment called a centrifuge, which spins rapidly in order to separate components of liquids based on their density. So, for example, it's possible to spin all the cells that contain DNA out of a saliva sample, or separate the white blood cells out of a blood sample. The liquid that's left behind looks like a valid biological sample, but contains none of the original DNA.


All that's left then is to replace the original DNA. Since most of the forensic tests are performed using a standard set of DNA fragments amplified by PCR, the authors simply took an environmental sample ("blood, dry saliva stains on absorbent paper, skin scrapings, hair, and smoked cigarette butts were collected"), amplified up the same fragments, and then spiked the purified blood and saliva with this DNA. They estimate that a library of about 425 DNA fragments would be enough to fake a match to just about anyone in existing DNA databases—without any DNA from that individual ever being obtained.
Of course, if any forensic technician goes digging beyond the standard genetic markers, problems would become apparent pretty quickly. Still, the authors are ready for this eventuality. A technique has been around since 2001 that allows the entire human genome to be amplified, starting from a sample with fewer than 10 cells. Here, you'd definitely need a DNA sample from the individual you're trying to frame, but it wouldn't take much of one.


To show that their results were more than an intellectual curiosity, the authors shipped some fabricated samples off to a third-party DNA testing facility. As far as the facility was concerned, everything looked legitimate.
Fortunately, in identifying the problem, the researchers have come up with a solution. DNA inside human cells picks up a chemical modification called methylation; DNA amplified in a test tube doesn't. It's possible to determine whether or not a given stretch of DNA has been methylated using standard lab techniques, although these are a bit laborious and time-consuming, and it's the sort of technique that hasn't made its way into forensic training yet. Still, testing for methylation in a DNA sample should provide an important quality control on the sample—at least until biologists figure out how to apply methylation in a controlled manner.


It would be tempting to view this as an arms race between the sophistication of fabrication and forensic techniques, but the fact is that your garden variety criminal is unlikely to have the skills and equipment needed to pull a fake off. What it may do is undermine the general confidence in the use of DNA evidence. There's simply no way of knowing a priori which cases might have a molecular biologist with a vested interest, and therefore which samples need to be tested for tampering. It appears that we'll need to start testing methylation patterns in every case in order to retain confidence in the general techniques.



In the meantime, the most immediate effect of the paper is that it's almost certainly set off a spasm of scriptwriting among those responsible for TV whodunnits.
Forensic Science International: Genetics, 2009.  DOI: 10.1016/j.fsigen.2009.06.009                                                                             

[garbon]

Gattaca, duh.

[The Minsky Moment]

This is hardly a new or shocking revelation.  It happened to be the main defense in the most notorious criminal case since the Lindbergh baby.

[DontSayBanana]

This is hardly a new or shocking revelation.  It happened to be the main defense in the most notorious criminal case since the Lindbergh baby.

OJ?

[Eddie Teach]

I'd say Leopold & Loeb and the Manson family trump OJ in the notoriety department.

[derspiess]

I'd just like to say I know what PCR stands for

[DGuller]

Good.  Hopefully this development wil lead to convictions in many cold cases.

[PDH]

I hope this isn't just another whiney liberal attempt to have fewer minorities in prison.

[Malthus]

I hope this isn't just another whiney liberal attempt to have fewer minorities in prison.

Naw, it's an attempt to warn us all of the dangers of offending a vengeful biologist.

[PDH]

Naw, it's an attempt to warn us all of the dangers of offending a vengeful biologist.

See, that should be a given.  Anyone who has access to chemicals that could turn your piss blue needs to be given all due respect.