The genetics researcher who reviewed Sen. Elizabeth Warren's, D-Mass., DNA in an attempt to counter President Trump's "Pocahontas" jibes compared her samples with those from people in Colombia, Mexico, and Peru rather than Native Americans in the U.S.

Warren's DNA was reviewed at her request by Carlos Bustamante, a Stanford University genetics researcher, who compared it to reference samples and reported that he had found a Native American ancestor "approximately 8 generations" ago.

Bustamante, however, didn't compare Warren's DNA against Native Americans who live in the continental U.S., citing cultural reluctance to submit to DNA tests. Instead, he used recent samples from other countries whose populations presumably share a lineage during human settlement of the Americas about 15,000-25,000 years ago.

Warren's attempt to rebut the long-ridiculed claim, which was accompanied by a slickly produced television ad, appeared to be a move calculated to clear the way for a 2020 presidential bid. But it may backfire by adding confusion and controversy.

Some genealogical researchers point to limitations in both science and social meaning of the test results, which seem unlikely to prevent political attacks that Warren falsely claimed to be an ethnic minority to become a Harvard University professor.

[More: Cherokee Nation to Elizabeth Warren: Drop dead]

Academic skeptics point out the murky nature of DNA reference samples, and others argue that even if Warren has a native ancestor, that doesn't make her an American Indian in a cultural or political sense, or validate past claims of Cherokee ancestry.

Bustamante used DNA reference samples from the 1000 Genomes Project, an international effort that recorded samples from people around the world. He wrote that the strongest proof of indigenous ancestry could be found along Chromosome 10.

Warren's sample was compared against 37-person reference groups from Nigeria, China, Latin America, and Europe, and then contrasted against larger samples of white British and predominately European-descended Utahans. Overall, Bustamante found her DNA to be roughly 95 percent European.

“How accurate are these? There’s no validation test,” said Tufts University professor Sheldon Krimsky, who studies ancestry research. “There are methods of doing it, I suppose. You would have to get, for example, independent evidence of someone’s ancestry from records or their own information from families.”

Krimsky said he wants to know more about the error rate of sequencing, and expressed concern about Bustamante’s links to paid ancestry services. Bustamante worked from 2011-2017 as a scientific adviser to the company 23andMe and since 2011 has advised Ancestry.com, according to his CV.

“It’s a little conflict of interest if the scientist has been consulting,” Krimsky said. “The scientist rather than emphasizing the uncertainties would more likely discuss the probability that it’s correct.”

Data Bustamante used from the 1000 Genomes Project has a complicated backstory. The project sequenced genomes from about 2,500 people around the world. The data was crunched using principle component analysis, or PCA, to determine the ancestry of participants, and data sets are freely available to researchers.

The Native American data used for the Warren analysis appears to come from 37 Latin American people determined by PCA to have indigenous ancestry, though 12 appear to be admixed, said Paul Flicek, senior scientist and team leader for vertebrate genomics at the European Bioinformatics Institute.

"My assumption is that these 37 were the individuals who clearly had Native American DNA in this region," Flicek said.

Although complicated scientific analyses gauged probable ancestry, verifying the data would be nearly impossible for a third party because participants were anonymous.

"All the individuals in the 1000 Genomes project are anonymous," Flicek said. "To my knowledge, no one has contact information for them as the links between the individuals and their samples where broken. All we know is that the individuals were adults and were apparently healthy at the time that they contributed their samples. It is generally considered unethical research to attempt to deliberately reidentify anonymous research subjects."

Anthropological geneticist Deborah Bolnick of the University of Connecticut said data limitations could restrain genetic conclusions.

“The inferences are only going to be as good as that comparative database. If the comparative samples don’t represent the ancestral sources for a person, then some of the conclusions may be inaccurate,” said Bolnick, who is researching patterns of genetic variation among tribes that originally populated the Southeastern U.S.

“You are comparing a person’s DNA to the DNA of other people alive today and what their contemporary affiliations are. There is a presumption that the social connections of someone today are going to be representative of the social identities and community memberships that existed in the past. Because people move around in time, people’s social identities change from generation to generation even if the DNA doesn’t change that much,” Bolnick said.

Ancestry testing is a significant business in the U.S., boosted by television commercials hawking surprise results exploding assumptions about family histories.

"They are privileging non-indigenous definitions of what it means to be indigenous," said University of Alberta anthropologist Kim TallBear, who believes Warren "owes an in-depth apology to the Cherokee Nation" for once claiming tribal descent.

TallBear is not a geneticist, but has closely followed scientific research, and said many present-day Native Americans define themselves by cultural affiliation or political tribal membership.

"It does show she probably does have an ancestor who was indigenous based on our identity today, but these markers are found in other parts of the world," TallBear said.

Even the genealogists who argue that there's real science behind the research and vouched for Bustamante's reputation drew attention to the limits of his report.

University of Kansas anthropology and genetics professor Jennifer Raff said "it's absolutely true that we don't know a ton about the genetic diversity of North American indigenous populations, due to a very understandable reluctance to engage in genetic research."

But she added that "we do have a pretty good idea of what at least some of the genetic variation shared by all indigenous peoples of the Americas looks like, and the methods that Carlos Bustamante used were designed to be used in this particular situation" and "while this isn't an ideal test, the conclusions are reasonable as long as they don't over interpret them."

Evolutionary genetics professor Mark Thomas of University College London said: "Bustamante is a highly credible and respected scientist." He added that the new research "seems to support Native American ancestry 6-10 generations ago, but 6-10 generations ago she would have had up to around 1,000 ancestors, many of whom she will not have inherited any DNA from."

Biostatistician Jay Kaufman of McGill University in Canada said that many "white" Americans may be much more Native American than Warren: "It is worth perhaps asking the question: how many Americans who self-identify as 'white' have similar or larger proportions of non-European ancestry?"