AUBURN — A Sabattus man convicted of murdering his ex-girlfriend in 2005 is seeking a new trial based on new DNA analysis techniques.

Androscoggin County Sgt. Martin Fournier removes the chain around the waist of Daniel Roberts in 2015 at 8th District Court in Lewiston for a post-conviction review. Roberts was back in court Tuesday seeking a new trial.  Russ Dillingham/Sun Journal file

Daniel Roberts, 54, appeared Tuesday in Androscoggin County Superior Court.

Roberts is seeking to have Justice Harold Stewart II reconsider an order issued by now-retired Justice Joyce Wheeler, who presided over his 2007 jury trial.

Wheeler denied Roberts’ post-conviction DNA motion for a new trial based on what, in 2019, was considered new evidence derived from the latest DNA extraction techniques.

A Smith & Wesson .38-caliber revolver found at the crime scene — Roberts’ Sabattus garage — had been tested for DNA evidence.

Roberts claimed Melissa Mendoza, 29, his ex-girlfriend and the mother of their 2-year-old daughter, had arrived at his home in the middle of the night holding the gun. He said he shot her in self-defense.


He was convicted and sentenced to 55 years in prison.

DNA found on the gun was examined in 2005 and again in 2018, with updated lab techniques.

In Roberts’ 2019 petition for post-conviction review, he claimed Mendoza could not be ruled out as one of three contributors to the mixture of DNA that was found on the butt of the gun.

Through his attorneys, Roberts asserted that prosecutors at trial introduced evidence that Mendoza had not handled the gun and that evidence had helped sway the jury to convict Roberts.

Wheeler wrote in her 2019 decision that the newly offered DNA evidence only proved that Mendoza “could have touched the gun. The conclusions go no further. There is no evidence to show that Mendoza was in fact the person who touched the gun; she touched it before she was killed; she intended — or even threatened — to kill Roberts or Savanna (their daughter); or that the jury would have been persuaded in the slightest by this evidence.”

Wheeler wrote that she “cannot say that this new testing in anyway confirms or even gives support to the petitioner’s theory” that Mendoza entered Roberts’ home with the gun and a first responding officer touched the gun when he moved it.


“At most, the court can only say that it does not rule out what the petitioner claims,” Wheeler concluded. “This, however, is not enough to make it probable that a different verdict would result upon a new trial.”

More recently, Roberts consulted with a so-called “probabilistic genotype expert,” whose analysis of the genetic material found on the gun goes beyond the traditional method of including or excluding the possible people whose DNA was on the gun.

Jami Harman, an FBI-trained DNA analyst who reviews case files and evaluates data from biological testing labs, testified Tuesday for the defense.

Her husband, Kent, created a software program with algorithms for forensic DNA interpretation for probabilistic genotyping, she said.

That means, instead of simply including or excluding someone as a possible contributor to a DNA sample as earlier testing did, probabilistic genotyping can put a number to the likelihood that someone was a contributor to a mixture of DNA profiles found on evidence.

When examining DNA material that has more than one profile or where there is less DNA material to test or the results are less clear, probabilistic genotyping can help determine to what extent it’s likely a particular person’s DNA was present.


In the case of the Smith & Wesson revolver, Harman said “at least four” different DNA profiles were found on the grip of the gun.

And it’s 7,030 times more likely that, within that mixture of DNA profiles, that Mendoza and three unidentified people contributed to the mixture than not her and simply four unidentified contributors to the mixture of DNA, Harman said.

“We are not stating that a specific profile was expected to occur within X-number of people,” Harman said. “We’re taking the profile as a whole and saying, ‘this is how much more likely it is that a specific person is present than not present under the stated conditions.'”

By contrast, Harman said there was “significantly more” DNA data present on the gun “that would be consistent” with Roberts.

Assistant Attorney General Donald Macomber pointed out that the probability number for Roberts, using the same method, was determined to be 3 septillion, 606 sextillion times more likely that it is his DNA on the same gun plus three unknown people than not him among four unknown people.

“I would consider it to be a very robust conclusion that he is more likely to be on that gun,” Macomber said.


Harman agreed.

“Especially when compared to 7,000, correct?”

“Yes,” she said.

The judge asked the prosecution and defense to submit their written arguments.

Related Headlines

Comments are not available on this story.