Highly effective DNA software program utilized in a whole bunch of prison circumstances is re-examined

DNA has been considered the gold standard for forensic evidence for more than 30 years, even though various types of junk science are no longer used. In recent years, police and crime laboratories have expanded and expanded the use of genetic material to identify suspects – from using private ancestral websites to creating police sketches with faces of suspects.

The latest method under scrutiny is an obscure technique called “probabilistic genotyping,” in which incomplete or otherwise unfathomable DNA left behind at a crime scene, often in tiny amounts, is passed through a software program that calculates how likely this is to have come from a particular person. One such program, TrueAllele, has been used in more than 850 criminal cases over the past 20 years. The problem? Nobody knows if it works – the code developed by a private company called Cybergenetics is proprietary.

Government crime laboratories using the software will not be given access to the program’s source code. Cybergenetics employees are not granted access. Even the authors of TrueAllele’s peer-reviewed studies have never had access to the code.

But now two criminal cases – one in the U.S. District Court for the Western District of Pennsylvania and one in the Appeals Department of the New Jersey Supreme Court – can give the world its first glimpse into TrueAllele’s secret algorithm. Last month, the New Jersey judge ordered prosecutors to hand over the source code for TrueAllele, and a few weeks later the Pennsylvania federal judge did the same.

Experts say the program is so complex and so hidden that software bugs are inevitable.

“It is practically certain that the TrueAllele software has defects,” wrote Mats Heimdahl and Jeanna Matthews, two computer science experts, in a statement to the federal court. “On average, there are six errors per 1,000 lines of code, and TrueAllele has 170,000 lines of code.”

While the contracts do not make the source code publicly available, they do allow the defense to call in experts to examine the software for possible errors and inconsistencies under strict nondisclosure agreements. Should the experts find problems, the defense lawyers could attempt to ditch the DNA evidence, weakening the cases against their clients and potentially leading to a bigger settlement for TrueAllele.

[Read: How do you build a pet-friendly gadget? We asked experts and animal owners]

How the software program works is far more complex and difficult to reproduce than the traditional process of DNA testing, which assigns a suspect to a robust type of evidence such as blood.

TrueAllele does not compare a suspect to physical evidence. It calculates the statistical probability that a person’s DNA is present in a complicated mixture of DNA from several people or in a tiny amount of DNA left behind – for example, after someone just touched something.

In the Pennsylvania case, prosecutors used the software to analyze an intricate mixture of DNA found on a pistol that was left in a car – and in particular to calculate the likelihood that any of it belonged to a Pittsburgh man named Lafon Ellis. Cybergenetics experts said the sample likely contained DNA from four people and that it was “21.4 trillion times more likely” that Ellis’ DNA was in the mix than that of any other random African-American person. However, Ellis’ attorney says the gun is not Ellis and denies the DNA evidence linking him to it.

In addition to prosecutors, defense teams have also used the software to prove their clients’ innocence in court or to exonerate them later. Cybergenetics has definitely made the program available to both sides for testing.

According to Cybergenetics, TrueAllele’s calculations have already been approved as evidence in 14 states, with 20 unsuccessful attempts by criminal defense teams to gain access to the program’s source code in the past few years. Prosecutors have long argued that the program is based on generally accepted mathematical concepts and that opening the source code to public scrutiny would compromise the company’s trade secrets.

The lack of control of the software

However, experts and civil rights activists have long been concerned about the lack of control over the software. Their arguments could gain in importance, as last month’s two judgments show.

“Our judicial system cannot accept convictions based on classified evidence,” the Electronic Frontier Foundation and the American Civil Liberties Union of Pennsylvania wrote in an amicus letter to the federal court. “There is a long history of junk science used under the guise of technological advancement in criminal matters – and public access to and analysis of such evidence as a means of invalidating it.”

From bite mark samples to blood spatter analysis to ballistic testing and fingerprint matching, many forensic methods that were once common in crime laboratories and courts were later questioned or even fully debunked after being subjected to external scrutiny.

Life or death software programs – such as those used in medical devices or airplanes – require independent validation and verification processes, but DNA software does not. There are few federal regulations about how police or crime laboratories adopt new technologies and methods to solve crimes. Each state has its own set of rules about how to allow a new type of forensic technology to be used, and judges tend to make earlier decisions when deciding whether or not to allow new scientific techniques as evidence.

Verifying TrueAllele is likely to be complicated – the inventor estimates that it would take a person reading 10 lines of code per hour about eight and a half years (which defenders deny) – and how to go about it was another source of friction between prosecutors and defense lawyers .

Originally, in both cases, the defense teams were surprised that their access to the source code was restricted to a read-only iPad and that they could only use a pen and paper to jot down. Since then, they have advocated an electronic copy that they can run and test for themselves.

While TrueAllele has never been tested like this before, two of its competitors have.

The competitors

When a federal judge in the US District Court for the southern borough of New York ordered source code access to New York’s proprietary DNA analysis software Forensic Statistical Tool (FST) in 2016, the investigation revealed a fatal error that “tends to”[ed] overestimating the likelihood of guilt. “The judge in this case eventually lifted the software protection order, and ProPublica received the code and posted it on GitHub for researchers and the general public to review for themselves.

The New York office of the Chief Medical Examiner replaced FST with STRmix, a widespread competitor of TrueAllele. So it happens that the makers of STRmix have also identified software bugs, which affected 60 cases in Australia.

“Software bugs are common and forensic software has no particular immunity to the bugs and errors that plague software in other areas,” wrote Khasha Attaran, the assistant federal defender who represents Ellis, in an emailed statement. “Constitutional principles and fairness require access because a defendant must be allowed to test and question the veracity of Software generated Evidence on which the federal government relies. “

The federal prosecutor’s office has the option to appeal the decision to the U.S. Third Circuit Court of Appeals, but has not yet indicated whether or not they will. The New Jersey case can go to the state’s Supreme Court. The U.S. Attorney’s Office, which was prosecuting the Ellis case for the Western District of Pennsylvania, and the Hudson County Attorney’s Office, which was pending the case in New Jersey, declined to comment on ongoing cases, as did Mark Perlin, who Founder of Cybergenetics and inventor of TrueAllele.

This article was originally published on The Markup and republished under the Creative Commons Attribution-NonCommercial-NoDerivatives License.