DNA tests are not always foolproof

Friday, Nov. 3, 2017 1:28 AM

DNA testing procedures developed by the medical examiner’s office in New York City have come under attack, calling into question the validity of some DNA-based convictions. The office recently stopped using two disputed procedures.

The New York Times quoted a joint statement from the Legal Aid Society and the Federal Defenders of New York that said the medical examiner’s office “has engaged in negligent conduct that undermines the integrity of its forensic DNA testing and analysis.” Because of “unreliable testing” and “unsound statistical evidence” concealed from the public and the courts, “innocent people may be wrongly convicted, and people guilty of serious crimes may go free.”

Timothy Kupferschmid, the medical examiner’s chief of laboratories, defended his office’s methods. He said the discontinued techniques were well-tested and valid and compared the adoption of replacement procedures to a vehicle upgrade.

DNA analysis is touted as the gold standard for forensic testing, and many members of the public believe it’s all but foolproof. It’s not uncommon for defendants to give up and plead guilty when told that the prosecutor has DNA evidence against them.

When correctly done, standard analysis of pure samples of blood or semen that may contain billions of copies of one person’s DNA is about as foolproof as scientific testing gets. The problem is that scientists have pushed the envelope in an effort to analyze the miniscule amount of DNA present in as few as three cells. They’ve also developed software programs that are used to calculate the probability that a given suspect’s DNA is present in a complex sample that includes DNA from several sources.

When only a tiny amount of DNA is available, enzymes are used to copy the DNA repeatedly until the sample is large enough to test. But every time a thing is copied, imperfections are amplified and errors creep in. The technique developed by the medical examiner’s office in New York increased the number of DNA copies beyond the standards of other laboratories. Hence, critics say, more imperfections.

The other questioned method uses a software program that considers the amount of DNA in a mixed sample, the number of people who probably contributed DNA to it, the amount of DNA information likely to be missing or contaminated, and the variable frequency of specific DNA sequences in different racial and ethnic groups. From this information, the program calculates the likelihood that the DNA of a given suspect is part of the mix.

The New York Times quoted several experts who say the medical examiner’s software program made assumptions that aren’t valid and that its statistical methods are indefensible.

Developers of the software acknowledged a 30 percent margin of error in the method of quantifying the amount of DNA in a sample. They also said the program didn’t consider the possibility that components of a DNA mixture could derive from closely related people or members of an insulated population such as Hasidic Jews who share ancestors and DNA sequences.

Defense experts in several court cases were denied access to the software source code on the grounds that the code was a “proprietary and copyrighted” tool owned by the city of New York. The medical examiner’s office denied a public information request from the nonprofit news organization ProPublica, citing the code’s “sensitive nature.”

Perhaps there’s a good reason for being cagey about the computer code underlying a now-discontinued testing method formerly used to develop evidence in criminal cases, but I can’t think what it could be.

Dr. Carol J. Huser, a forensic pathologist, served as La Plata County coroner from 2003-12. She now lives in Florida and Maryland. Email her at