'Game-changer' DNA software put to the test in Minnesota

In a small, quiet courtroom in Anoka County, forensic software designed to interpret tricky pieces of DNA evidence recently confronted what's believed to be its first test in Minnesota.

Proponents of the software tout its ability to help scientists examine DNA mixtures previously considered too complex to untangle. Such mixtures from multiple people may arise in mingled bloodstains at a crime scene, for instance, or from a stew of skin cells from "touch DNA" swabbed from a surface like a doorknob.

STRmix (pronounced star-mix) and other software like it rely on computers to interpret these mixtures. Such programs mark a crucial shift away from other approaches, which have come under fire for being too subjective and vulnerable to human error or bias.

"I don't claim complete objectivity," said John Buckleton of the New Zealand Institute of Environmental Science and Research, one of the developers of STRmix. "There are elements of subjectivity remaining, but it's a significant move toward objectivity."

The recent Anoka County hearing represented a first hurdle on the leading edge of this software's use in Minnesota, with judges, defense attorneys and prosecutors expected to weigh its use in court in the coming months.

Forensic scientist Anne Ciecko took the stand Tuesday to testify about the mathematical underpinnings of STRmix, explaining how the software helps unlock more of the data found in a DNA profile.

"It accounts for things we couldn't account for before," Ciecko said. "It helps us do the math."

Her testimony was part of what's called a Frye-Mack hearing, used to determine if scientific evidence tied to a "novel" process or technique can be admitted in court. District Judge Dyanna Street now will decide whether to allow STRmix evidence in a Fridley homicide case, described in court documents as a drug deal gone bad. Street's favorable remarks at the Tuesday hearing bode well for the software's admissibility in the coming trial.

STRmix is making its way into criminal investigations after the Midwest Regional Forensic Laboratory (formerly known as the Tri County Regional lab) in Andover went online with the software in January — the first lab in the state to do so. The regional lab, where Ciecko works, joins nearly 30 forensic labs across the country, including at the Federal Bureau of Investigation, in using STRmix. Dozens more, including the Hennepin County Sheriff's Office, also are working on getting the software up and running.

What once took hours now often takes mere minutes with STRmix, scientists say. Labs can tackle more mixtures and tease out more findings in a fraction of the time.

Long hailed as a gold standard in forensics, DNA testing has been called upon in all types of crime, from homicide to vandalism. It's been used to identify and convict suspects as well as to free scores of innocent prisoners.

Though police say DNA isn't the be-all and end-all of an investigation, it often provides crucial support.

That's especially true of samples from one individual or simple DNA mixtures from two individuals (as in sexual assault cases), samples that are considered reliable and largely objective.

The more people in a sample, however, the trickier the DNA analysis becomes, especially if there's not much DNA in the first place. Subjective interpretations of complex mixtures by scientists have proved especially problematic in recent years, as indicated in a 2016 federal report to the president's office on forensic science.

Enter STRmix and similar software, which rely on "probabilistic genotyping" programs to report the statistical likelihood that particular suspects contributed their genetic material to a mixture of evidence.

These statistics, called likelihood ratios, have changed the way scientists report some of their findings to police and prosecutors.

Scientists at the Midwest Regional lab, for instance, previously may have concluded that DNA swabbed from an object excluded 99.98 percent of the population — but not a particular suspect. With STRmix, the report instead pairs a ratio with a verbal equivalent, such as "extremely strong support," "moderate support" or "weak support" that a suspect's DNA was part of the mix.

"I know there was some concern early on with the prosecutors," said Cmdr. Bryon Fuerst of the Anoka County Sheriff's Office. "They have to explain this to a jury."

Fuerst said grasping 99 percent with a string of 9's after the decimal point is easier "to wrap your brain around" than clunky probabilities.

Prosecutors have to be cautious about how they present likelihood ratios in court, said Paul Young, criminal division chief at the Anoka County attorney's office.

Young, a prosecutor in the Fridley homicide case, described STRmix as a "game-changer" for its potential to improve results from DNA mixtures.

In the Fridley case, two men have been charged with second-degree murder in the May shooting death of James Chapman, 54, of Minneapolis. A gun provided what could be crucial evidence in the case involving Johnny Earl Edwards, 44, of Brooklyn Center, who's one of the defendants.

Though STRmix is only now making its way into Minnesota, it has been used elsewhere since 2012. Supporters say STRmix has been validated, with its methods scrutinized by scientific groups even as efforts grow to assess how DNA mixtures are interpreted.

"We can never say with any DNA result, 'Hey, based on this finding, that means this is so-and-so's DNA,' " Young said. "All we can do is present the statistic to the jury. They are the only ones who can ultimately make the conclusion."