By Lee Dryden
The Daily Record Newswire
DETROIT - A 2014 armed robbery in Muskegon County led to what is being hailed as a groundbreaking decision for the way in which DNA evidence was used to help convict the suspect.
The conviction is believed to be the first in the nation based in part on a brand of software used for probabilistic genotyping after a defense objection. It's an automated program used to analyze DNA when samples from more than one person are mixed. It has been admitted in New York cases when unchallenged by the defense.
Elamin Muhammad was convicted of armed robbery March 18 after a bench trial by Muskegon County Circuit Judge William C. Marietti. Central to the case is Muhammad's shoe that came off when he fled while dodging bullets from the gas station clerk during the robbery.
The shoe contained DNA from more than one person and the Michigan State Police crime lab was unable to narrow it down. Prosecutors eventually pursued the probabilistic genotyping approach through STRmix software that showed Muhammad to be the major donor.
Muskegon County Senior Assistant Prosecutor Robert F. Hedges said it is a legitimate tool that has prompted calls to his office from prosecutors in other states hoping to learn more.
"It's gaining popularity all over the place," he said. "It will help police solve cases that were previously unsolvable."
Scott E. Pederson, who represented Muhammad, objected to the evidence and said there will be an appeal, although he won't be handling it.
"I think the judge wrongly used it, relied on it," he said.
What is probabilistic genotyping?
Probabilistic genotyping is practiced via at least two types of software: STRmix, as used in Muhammad's case, and TrueAllele, whose developer Mark Perlin is considered a pioneer in the field. STRmix expert and New Zealand native John Buckleton testified in the Muskegon County case.
"It's not a substitute for DNA collection or amplification," said Erica Beecher-Monas, a law professor at Wayne State University Law School. "Law enforcement still collects the samples and submits it to the lab, which amplifies the DNA in the sample that can be statistically analyzed."
The program - as opposed to a person - analyzes the samples and applies algorithms comparing different statistical models to the data, Beecher-Monas said.
"It is now being applied to low quantities of mixed DNA samples with multiple contributors," she said. "Low quantities of mixed sample DNA - typical in crime scenes - have been a long-standing problem in DNA analysis."
There are consequences from trying to manually interpret complicated DNA mixtures from two or more people, Perlin and Kevin W.P. Miller wrote in a 2014 article in Forensic Magazine that discussed early adopters of the practice.
"Hundreds of thousands of evidentiary items have been collected and processed into DNA identification data that could implicate or exonerate - but these items have not been reported conclusively," they wrote. "A failure to fully use data from evidence is a failure of science to promote justice. This ongoing mixture crisis permits needless victimization by those whom DNA should have identified."
The Michigan State Police crime lab began using STRmix in March with a limited number of users, said Shanon Banner, manager of the state police's public affairs section.
This software standardizes the department's interpretations of mixtures with hopes it will lead to fewer inconclusive results, Banner said.
The Michigan case is being hailed as the first to successfully use STRmix amid a defense challenge. TrueAllele evidence has been admitted in several states, including Ohio and Pennsylvania.
Is the evidence reliable?
In a December opinion allowing expert testimony about STRmix evidence, Marietti wrote that the concept has proven to be reliable and accepted among experts in many fields. He cited the admission of similar evidence in other states.
After the Michigan State Police lab analysis found multiple donors, prosecutors sent it to a Pennsylvania lab that found a mixture of two people with a match to the defendant, Hedges said. The Pennsylvania expert withdrew the opinion after a defense objection and recommended the probabilistic genotyping route.
While awareness is growing in the legal realm, the practice isn't new, Hedges said.
"This just didn't appear out of nowhere overnight," he said. "Given that history, I fully expected we would prevail on the admissibility."
Hedges said Buckleton traveled to Michigan to testify on STRmix and testified via phone on another occasion.
The prosecutor said genotyping can be used for convicting and exonerating people - such as with innocence projects - but acknowledged defense attorneys likely have a different viewpoint.
His assessment is correct.
"I see this as a pure police tool," Pederson said. "I don't think the DNA evidence should have been part of the trial - at all."
Perlin found via the genotyping method that a female was the major DNA donor in Muhammad's case, Pederson said.
Despite his objection, Pederson acknowledged the significance of the Michigan case.
"The judge found it reliable and he used it in his decision," he said. "It's a revolution that started here in Michigan."
Muhammad was very active in his defense and involved in research.
"He will definitely appeal," Pederson said.
STRmix's website touts it as "a breakthrough for forensic analysts as it can assist investigations using DNA evidence that was previously considered too complex to interpret."
It states the program uses more of the information in a DNA profile, interprets results faster, compares profiles against a person of interest and helps with cases where there are no suspects but multiple DNA contributors in one sample.
Beecher-Monas said it is difficult to determine if the results are legitimate as it is a proprietary system and there is a risk of human error and data manipulation.
"TrueAllele software requires making assumptions about the number of contributors to the DNA profile, which may affect the accuracy of the results," she said. "The TrueAllele system relies on random statistical modeling, in which the randomization varies with every test.
"Should TrueAllele open its source to independent investigators for testing? Yes. Do the courts care? Probably not," she said.
Prosecutors and defense could use the software if they purchase it, but it is unclear if the untested system would be helpful, Beecher-Monas said.
"I suppose the defense could run their own analysis, but if the results differ, who is to say which analysis is correct? Without testing we can't know," she said.
Caitlin Plummer, a clinical fellow at the Michigan Innocence Clinic at University of Michigan Law School, agreed genotyping could be helpful to prosecutors and defense attorneys, but also warned the complex technology should be validated and tested outside of the courtroom to ensure accuracy.
"Too often the criminal justice system has accepted forensic techniques that later turned out to be unreliable - meaning innocent people were convicted, and the true perpetrators escaped justice," she said. "And the system probably could have avoided those injustices had more care been taken at the outset to make sure the science was valid."
If previously unusable data can be used reliably, that is "big news for the criminal justice system," Plummer said.
"But reliability is the million dollar question," she said. "For better or worse, the courts are leading that inquiry right now. Only a handful have been confronted with this technology thus far, and I think it's too soon to tell which direction things are headed."
Plummer added that this approach does not show how or when DNA was deposited at a crime scene and caution should be taken to consider possibilities such as contamination and transfer.
Published: Tue, Apr 12, 2016
