May 11, 2006

It's All in the Family

So DNA fingerprinting isn't especially effective at identifying criminals. There's that whole "no such thing as race" problem. And the "we don't really have a database of people of asian descent" thing. Oh yes and the problem of getting a good sample. Anybody who has ever seen CSI - every bit as educational as biology class - knows that forensic genetics is, um, still pretty medieval in terms of identifying subjects. It can rule them out. But plenty of folks have been falsely convicted because juries have ridiculously exaggerated faith in DNA based evidence. What to do?

Our pal Jed Gross at Yale points us to this brilliant solution:

Cops would solve more crimes if they expanded their use of the nation's DNA fingerprinting system to test close relatives of known criminals, according to a research report that raises novel and difficult civil liberties issues.

The proposed crime-control strategy, already in growing use in England, is based on two central facts: Close relatives of criminals are more likely than others to break the law, research has shown, and, because those individuals are related, their DNA "fingerprints" will be similar. That suggests that if police find a crime-scene specimen with a DNA pattern close to -- but not exactly the same as -- that of a known lawbreaker, a relative of that known criminal may be the culprit.

And the thing is, in America, lots and lots of people who commit crimes are related to others who also have done so:
In the United States, those odds are rather high: A 1999 Justice Department survey found that 46 percent of jail inmates had at least one sibling, parent or child who had been incarcerated at some point. All states take DNA from all convicted felons, and many get specimens from a wide range of others. Using conservative assumptions, Bieber and his colleagues calculated that U.S. law enforcement authorities could increase their "cold hit" rate (the percentage of DNA searches that result in perfect matches) by 40 percent if they were to check the DNA patterns of criminals' family members when searches generate near misses.

Cold-hit rates vary widely today. Assuming they average about 10 percent, Bieber said, a 40 percent increase would bump that rate up to 14 percent.

"This is proof of concept that existing methods of kin analysis could be used in forensic analyses with an expectation of success in a fair proportion of cases," said Bieber, who is to present the findings tomorrow at a meeting of the American Society of Law, Medicine & Ethics in Boston.

View blog reactions

| More