DNA Software Use in Criminal Investigation Growing

By Shuangming Pang


New software is helping police track down criminals with absolute precision, but civil libertarians argue that the algorithm used to make its matches must be made public.

One case in which the technology resulted in a conviction was in the summer of 2013 when three women were reporting sexually assaulted in Bakersfield, California. The victims generally identified their attacker as a man wearing a ski mask. Police believed a single perpetrator was responsible for these crimes. The serial rapist was found guilty and sentenced to life in prison for 423 years in 2015.

Kern County, California law enforcement linked Billy Ray Johnson to the rapes by DNA and clothing evidence. Johnson had a record of theft, assault, and domestic battery. The process of determining the criminal’s DNA was complicated: the DNA samples collected from blood stains and touch DNA were too delicate, and there was a mix of DNA from different people.

It was a scientific method called Probabilistic Genotyping that helped the police to crack down the criminal. It uses mathematical formulas to interpret complex DNA evidence that contains genetic material from multiple people. The Kern County Crime Lab used a probabilistic genotyping software called TrueAllele, sold by Cybergenetics of Pittsburgh, Pennsylvania.

DNA Image (Courtesy of ACLU)

However, two years after Johnson’s sentence, the American Civil Liberties Union of Southern California filed an amicus brief to the California 5th District Court of Appeals in the case called California v. Johnson on Sep 14, 2017.

The brief argues that the use of the process violates Johnson’s constitutional rights and “the complete lack of transparency as to an algorithm that is material to a criminal trial vitiates the public’s First Amendment right of access,” wrote Peter Bibring, an ACLU senior staff attorney.

The ACLU is requesting that the court order Cybergenetics to release the source code that runs the probabilistic genotyping software to execute the algorithm. They argued that the algorithms are “human constructs that include numerous sources for bias and mistake,” Bibring wrote in the brief.

The ACLU is trying to raise the possibility that the human-developed DNA software is not perfect, although DNA evidence that is supported by this type of software in court cases is generally seen to be indisputable.

“These tests run by private corporations are going to be used to put people into cages to deprive their freedom,” said Kade Crockford, the director of technology and liberty program at ACLU of Massachusetts. “The subject of the investigation should have the right to have access to the algorithm and all kinds of other information about the scientific means by which the company arrives its conclusion.”

After an allegation from an ACLU blog, which said the source code is a secret, the Cybergenetics response in its website said that TrueAllele computer methods are not a secret. “Anyone can test the software by running data through the system,” said Cybergenetics in the post, adding, “it also offers free cloud-based testing to defendants.”

For the validation argument, Cybergenetics site stated said they had been published in scientific journals, with more information provided in criminal proceedings.

The forensic community is switching the general process of searching from random math probabilistic statistics to widely issue-based statistics using probabilistic genotype software. STRmix, a probabilistic genotyping software created by scientists in New Zealand and Australia, also faced the validation challenges by civil libertarians.

The ACLU filed another brief in California v. Dominguez in July. A man called Florencio Jose Dominguez was accused of the murder of 15-year-old Moises Lopez in 2008. The San Diego crime lab used STRmix connecting Dominguez to the crime. However, in October 2017, a California trial judge ordered that the state disclose to the defense information about STRmix.

“I believe the Johnson case has not been heard yet,” Vera Eidelman at ACLU’s Speech, Privacy, and Technology Project said. “It is an appeal of a trial that already happened.”  The Dominguez case is scheduled for oral argument on Sept. 13. It is likely to move faster because the appeal happened before the trial occurred, Eidelman said.

STRmix is the most widely used DNA software in the U.S. By June 2018, STRmix was in use in 34 U.S. crime labs, all  nine state and territory labs in Australia, and seven labs elsewhere, said John Buckleton, a forensic scientist and one of the developers of STRmix.

States that Use STRmix DNA Analyze Software 2014-2018

New measurability challenges are coming out. Civil libertarians were trying to get the source code for the sophisticated software.

Usually companies would work out a non-disclosure type of agreement, explained Carll Ladd, a DNA scientist at the Connecticut Crime Lab. “so they can review the source code, but they cannot give it away to the third party to protect those companies’ business interest.”

“Probabilistic genotyping has to go through this process like any new methodology would,” Ladd said, “I am anticipating that the court found it sufficiently reliable and accept it.”

Connecticut has been using STRmix since March 2017. The STRmix significantly improves criminal investigations because it can interpret complex DNA results, Ladd said.

It’s relatively early in the process for criminal investigation because it could interpret complex DNA result, Ladd said. It’s also relatively early in the process of the courts reviewing probabilistic genotyping software. Probabilistic Genotyping is a very complicated science method, and it is hard for the court and prosecutor to understand,  Ladd said.

“The argument today for the Probabilistic Genotype matching is the same argument that we used for DNA testing in general,” Ladd said. “It’s super complex, the courts, the juries will never get it.” He added, “In the long run, it will give greater confidence in the result that can be presented in court.”

About Shuangming Pang 4 Articles
Shuangming Pang is a multimedia journalist. Shuangming got the Bachelor degree in journalism at Shanghai University of Sports. She covered the 2016 Rio Summer Olympics as an intern producer of the sports channel at Shanghai Media Group, as well as a news editor of the college sports magazine during her undergraduate studies.