The Oklahoma Court of Criminal Appeals has unanimously affirmed the use of BulletProof, a probabilistic genotyping software program, in the assault and battery trial of Patrick Marquise Napoleon, who was convicted for the 2017 stabbing of Nancy Jones outside Tulsa’s Hillcrest Medical Center. The published decision opens the door for prosecutors in Oklahoma to utilize probabilistic genotyping software programs, called PGSPs, during criminal trials.
Used to analyze complex or mixed DNA samples, PGSPs calculate likelihood ratios for who might be a contributor of genetic material. The algorithms generate a probability that “the evidence gathered would have led to the evidence profile that was obtained, if the [person of interest] were — or were not — a contributor to the sample,” according to the U.S. Government Accountability Office.
In his opinion for the state’s top criminal court, Judge Scott Rowland held that testimony from an expert who analyzed the DNA sample in Napoleon’s case “shows BulletProof meets the Daubert factors,” a framework by which courts assess the reliability of expert witness testimony.
“When the evidence presented is weighed against the relevant Daubert reliability factors, we, like the district court, find that BulletProof’s DNA identification methodology and the match statistic reported in this case was sufficiently reliable to warrant admission,” Rowland wrote. “Accordingly, we find the district court did not err in admitting the PGSP evidence.”
The first attempt to prosecute Napoleon for the crime ended in a mistrial. To start Napoleon’s second trial, Tulsa County District Judge William LaFortune held a pre-trial hearing to determine the admissibility of the PGSP evidence under the U.S. Supreme Court’s 1993 decision Daubert v. Merrell Dow Pharmaceuticals. LaFortune heard testimony from Kent Harman, the CEO of Sorenson Forensics at the time, who argued that PGSPs are used in other states’ courts and by the FBI and ATF. As a result, LaFortune allowed the evidence to be admitted.
The Dec. 18 appellate decision affirms Napoleon’s conviction, which received a sentence for 65 years of imprisonment.
PGSPs deal with DNA ‘mixtures’ of two or more people
On April 6, 2017, Jones was returning to Hillcrest Medical Center from the QuikTrip across the street “before nightfall” when she was stabbed in the neck from behind. Her assailant fled to the nearby parking garage while Jones was rushed inside the hospital. In the parking garage, police recovered a bloody knife, camouflage pants and a white sock that they believed “the attacker wore over his hand.” Napoleon was arrested nearby within 15 minutes of the incident, and two hospital employees identified him as the attacker.
An initial DNA test run by the Tulsa Police Department’s forensic lab “could neither exclude Napoleon nor include him because the DNA collected from both items was a mixture of two or more persons.” Prior to their second trial against Napoleon, prosecutors enlisted Sorenson Forensics to run “probabilistic genotyping testing via the software program BulletProof.”
Sorenson Forensic’s new analysis of the test results, paid for by the state, found that “it was far more likely that Napoleon’s DNA was a major contributor to the mixture of DNA on both the sock and pants and that his DNA exceeded 50 percent of both mixtures.” With the newfound PGSP evidence admitted for his second trial, the jury convicted Napoleon.
On appeal, Napoleon argued that PGSPs are not reliable and that his defense team was not provided the source code for BulletProof. The court found his arguments unconvincing.
“Napoleon primarily challenges the reliability of the DNA identification evidence generated by BulletProof. He contends that the state never provided him with BulletProof’s source code information and that without it, there was ‘no way for the trial court to determine that such evidence was reliable or based on sufficient facts or data,'” Rowland wrote. “This claim is not supported by the record because, according to the expert, BulletProof uses EuroForMix algorithms to perform its mathematical calculations and its code is ‘open source[,]’ free, and downloadable for review of its accuracy, as well as being peer reviewed and generally accepted in the scientific community.”
Napoleon had also objected to the admission of the DNA evidence by alleging members of the jury at his first trial handled the sock and pants without gloves, thus contaminating the evidence. The court disagreed, finding PGSPs account for potential contamination.
“[Harman] explained that probabilistic genotyping accounts for multiple contributors and explained how it does so and how ‘contamination’ from others does not render its results unreliable,” Rowland wrote. “Napoleon offered nothing to contradict the expert’s conclusions and methodology.”
(Clarification: This article was updated at 11:25 a.m. Monday, Jan. 5, to clarify that Kent Harman is no longer the CEO of Sorenson Forensics.)
