Sign Up    Log In    Category    Coupons    Shops    Ideas    Reviews set favor Add to Favorites
 Home | Art & Crafts

Matching Science of DNA With Art of Identification
The type of test the U.S. government used to confirm it had killed Osama bin Laden has been widely used for decades in paternity disputes, criminal investigations and to identify remains after disasters. These genetic tests are well understood and generally reliable, but it can be tough to get a handle on what the results actually mean.

Preliminary DNA tests showed with 99.9% certainty that the corpse believed to be bin Laden's was in fact his, according to a U.S. official. Later tests were "absolutely conclusive," the official says. The U.S. hasn't released further details about the tests, but officials have cited other factors used in the identification: an unspecified facial-recognition test and intelligence pointing to his location in the raided compound in Pakistan.

Enlarge Image

Associated Press
Genetic-test experts say claimed match probabilities, such as 99% or 99.99%, can be misstated or misleading.

Genetic-testing experts say such additional evidence is key to ensuring a positive identification and they have no reason to doubt the conclusion in the bin Laden case. But they say probabilities such as 99% or 99.99%—frequently tossed around in other cases involving DNA testing—can be misstated or misleading.

"You could say, is it a big deal, does it matter?" says Bruce Weir, chairman of the biostatistics department at the University of Washington. "It does matter. It can lead to overturning of convictions."

Prof. Weir and other biostatisticians point to what is known as the "prosecutor's fallacy," the failure to make the distinction between the raw result of a genetic test, which is objective, and the probability that the identification is a match, which often is subjective.

The Numbers Guy Blog

DNA and Bin Laden's Positive ID
The problem boils down to this: A very small chance of a false positive in a genetic test isn't the same thing as a very large chance of a positive identification.

A genetic identification test involves comparing genetic markers from an unknown source—in bin Laden's case, the corpse believed to be his—with some known source, which could be a prior sample collected from the same person, or one from a close relative. Testers compare several parts of the genetic code, check where the samples match, and take into account how rare the genetic markers are.

This produces the probability of a coincidental positive test result, what some call a random-match probability. In other words, the test returns a probability that a random person would have a match as close or closer than the test finds.

The chance of a random false positive depends in part on what genetic investigators have to work with. When they have a sample from the person they are trying to identify and the test indicates a very strong match, the probability of a random false positive can be a very small number, on the order of one in billions or one in trillions, experts say.

Parents, children and siblings provide less certainty, because they may share some, but not all, genes. It is unclear what the U.S. used as a reference for bin Laden's genes.

The complication is that a very small chance that each person in the world other than a suspect would produce such a genetic match could add up to a decent overall chance of such a match with at least one person. For example, crime analysts have found dozens of partial matches among unrelated people in a DNA database of 65,000 Arizona felons.

"Unfortunately, and surprisingly to many people, the DNA test cannot identify the probability" of a positive match, says Jay Koehler, professor of law at Northwestern University.

That can have an impact in court. A federal appeals court in 2008 overturned the conviction of a Carlin, Nev., man convicted of rape because a witness for the prosecution confused the random-match probability with the chance the suspect had committed the crime.

The proper way to go from a match percentage to a probability of a positive identification is to incorporate what statisticians call prior odds. This notion, central to a discipline known as Bayesian statistics, refers to the odds that the person suspected of being a match is a match, irrespective of the DNA evidence. The higher the probability of a person being a match going into a test, the higher the probability of a positive identification after a positive genetic test.

There might be little reason for suspicion when comparing samples randomly in a large genetic database and so it might not be too surprising to come across a false match. The additional evidence in the bin Laden case supports the interpretation of a positive genetic match as a positive identification.

Prior odds are sometimes applied unevenly or arbitrarily. In paternity disputes, courts generally assume that there is a 50% prior probability that the possible father does indeed have paternity.

"The convention leads to absurd results" in some cases, says William C. Thompson, professor of criminology, law and society at the University of California, Irvine—for instance, when two men, often brothers, have positive genetic tests and therefore each is found to have a probability of paternity greater than 95%.

A forthcoming paper from University of North Texas researchers bemoans the lack of standards in computing prior odds, and suggests guidelines for missing-persons tests. But the prior-odds calculation will likely remain subjective in many cases.

So will the threshold for deciding when DNA evidence, coupled with prior knowledge, is enough to conclude a DNA test has produced a match. To confirm the identity of remains from the World Trade Center, investigators initially required that the probability of a random match be less than one in 10 billion. Later they loosened the restrictions slightly, because more remains had been identified so there was less risk of a false positive match.

And these probability calculations don't account for the possibility of a laboratory error. "The risk of human error far exceeds the risk of error due to a coincidental match," says Prof. Koehler.
 Jun 25 2012

previous: The Top Ten Car Brands
next: How to Choose the Right Gift

Same category shops:
we heart craft and stationery. We Heart Craft and Stationery. wholesale,diy,discount,sale,stationery,stationary,diy,ka...
uk's largest doll production company, we offer reborn & sculpting lessons.baby supplies, artist dolls, paints, brushes, ...
Wild Things Beads - Beads Wholesale to the Trade Wild Things Beads - Beads Wholesale to the Trade

fraud   Copyright © 2010-2014 ShopQuicker, Inc.  All rights reserved.      Contact Us    |   Tell Friends    |   Privacy    |   Terms of Use Link shopquicker  DeepData©