Thursday, 20 August 2009

Fabricated genetic fingerprints and the limits of forensic science


In CSI, the phenomenally successful television show about forensic science, DNA evidence is usually presented as something of a clincher. It's often said that a crime-scene sample that matches a suspect's genetic fingerprint leaves only a one in a million chance that he or she is innocent, and this sort of evidence is often among the most likely to convince judges and juries.

The police and the Home Office see it as so valuable and reliable that they want the Police National DNA Database to retain samples from people who are arrested but never convicted, even though the European Court of Human Rights has ruled this unlawful.

Since it was developed by Sir Alec Jeffreys in 1984, DNA fingerprinting has indeed helped to solve thousands of crimes, both convicting the guilty and exonerating the innocent. But valuable as it is, it is no more infallible than any other investigational tool. A fresh reason why has been highlighted this week by the New York Times, with a story suggesting that genetic evidence can easily be faked.

A team from an Israeli company called Nucleix, led by Dan Frumkin, has fabricated samples of blood and saliva that were provided by one person, but which carry the DNA of another. The scientists, whose work is published in the journal Forensic Science International: Genetics, also showed they can build DNA samples to match anybody whose genetic code is recorded in a database, without even obtaining any tissue.

The implications sound pretty worrying. An enterprising criminal could cover his tracks, and frame someone else, by collecting cells from a glass they had used and turning it into a saliva or blood sample carrying the innocent person's DNA. An unscrupulous policeman with access to the DNA database might even be able to use its contents to create evidence that would implicate a suspect. The techniques involved, Frumkin claimed, are not particularly complicated. "You can just engineer a crime scene," he told the NYT. "Any biology undergraduate could perform this."

It's quite true that DNA evidence is potentially open to manipulation in this manner. I'm not entirely convinced, though, that this paper undermines its value in criminal investigations quite as extensively as it might seem.

For a start, as The Register notes, it's already perfectly plausible for a criminal to plant hair from another person at a crime scene, without recourse to a laboratory. Frumkin's claims about the simplicity of making DNA evidence notwithstanding, few criminals will have the resources or know-how to do it. And in many cases, their own DNA will be more abundant than the evidence they plant -- a rapist, for example, will still be vulnerable from the semen he leaves behind him.

Frumkin's company, incidentally, is trying to sell police forces a system it has designed for telling manufactured DNA apart from the real thing, by analysing methylation patterns (which are involved in switching genes on and off). It clearly has an interest in promoting the idea that this is going to be a major problem.

What this research does highlight, however, is the danger of treating DNA evidence as something special. Important as it can be, it can only ever be one element of the evidence that builds a case beyond reasonable doubt -- and that applies whether or not it is easy to fake.

The one-in-a-million probability of a chance match is a good place to start. The basic maths are correct enough, but that doesn't mean that the chances that a suspect who matches that sample is innocent are also one in a million. All that a match does is to place that person at the crime scene -- and most crime scenes will be littered with DNA from perfectly innocent people. Absent further evidence pointing to guilt, that should never be sufficient to convict.

What matters is the context in which DNA is found, and the supporting evidence. DNA from semen found on a rape victim, or skin cells under the fingernails of a murder victim, is one thing. A few hairs on the floor of a corner shop that has been robbed are quite another.

However simple it is to fabricate DNA evidence, it should never be trivial to plant it in a properly convincing way. So long as we see genetic fingerprints as one tool among many for solving crimes, and do not exaggerate what they can tell us, we should be reasonably safe from miscarriages of justice.

Visit Times online for more information

1 comment: