Hi, Andrew. Just a cautionary note that much admixture estimation is, as Andrew noted uptopic, just that: a lot of guesstimation grounded in some baseline data, so conclusions have to be approached cautiously.
As one example, the Eurogenes project is, as you can guess by the name, focused mainly on European ancestry with sample cohorts selected accordingly. But those are also dated now: the Eurogenes Genetic Ancestry Project (
http://bga101.blogspot.com/) ceased updating datasets at GEDmatch about six years ago.
Too, the GEDmatch "Admixture Proportions by Chromosome" reports values as percentages, not centiMorgans. So under Eurogenes K13 you show 5.8% "Amerindian" on Chr 19, not 5.8cM. In other words, there is no implied segment continuity at all, just that of the 3,105 SNPs K13 looks at on Chr 19, 5.8% are reported as most closely aligned to what Eurogenes uses as an "Amerindian" reference sample. Since the centiMorgan calculation differs between the male and female genomes (cMs are based on a probability of recombination events, or crossovers, each time a gamete is formed; they aren't a physical measurement), using sex-averaged values for such small DNA chunks as admixture represents just doesn't work. For example, Chr 19 calculates out to about 96cM for males, but about 127cM for females.
Just an aside, but Chr 19 has the highest density of protein coding genes of any human chromosome; on a genes-per-base-pairs basis, the density is over twice the genome-wide average and takes up more than 25% of the entire chromosome. Some genes are just phenotypic--hair color, eye color, etc.--and those can be useful in population genetics, but the genes dealing with function and health are typically not useful at all because that's the DNA we have, pretty much, entirely in common.
Of the GEDmatch admixture calculators, a lot of folks, like Mercedes Brons, recommend the MDLP Project World22 calculator. It, too, hasn't been updated in a while, but you can read more about the datasets Vadim use and the sources from which he derives them at one of his 2012 blog entries. It offers an attempted breakdown into sub-continental populations more germane to Native Americans.
Truly, though, Native American autosomal DNA is very, very difficult to use for any genealogical purpose. It can certainly be used just like any other for recent generations, but very few Native Americans seem to have an interest in testing. For good reason. The nature of the Beringian diaspora and a possible southern immigration at approximately the same time meant that the relatively small founder populations in the Americas were essentially isolated for over 13,000 years; the new continental-level population spread from today's Alaska to the southernmost tip of Chile to Newfoundland to eastern Brazil. The people covered an area that's 28% of the world's landmass, an area 40% larger than Africa and about 4.25 times the size of all of Europe.
So for thousands of years the population had very little exogenous DNA introduced. Some hypotheses indicate that there were both northern and southern basal branches--the first from southern Siberia and the second from mainland Southeast Asia--and studies as recently as December 2019 (Pakstis, et al., Scientific Reports) indicate that, at most, we can identify nine groupings, genetically speaking, of all the founding populations in the Americas. So not only did the people begin to become much more admixed after the arrival of Europeans over 500 years ago, but still today someone having pre-Columbian Central American ancestry can be genetically indistinguishable--from the standpoint of origins--from one of the great Plains nations of Native Americans.
Admixture information can be interesting and, sometimes, even informative. But regardless of the source--whether one of the testing companies reporting "ethnicity" (a term I really wish the advertising companies hadn't talked them into) or one of the independently created tools at GEDmatch--it usually involves reference sets that are either individual or tiny clusters of SNPs, or segments way too small to be accurately usable for genealogy. And that's especially true for low or trace percentage amounts.
Finding out you show 10% East Asian admixture could be a big clue for genealogy research if it was unknown to you, but 1% "Amerindian" could as easily be the result of a 10g-grandfather who was a Spanish officer marrying an indigenous woman in Panama in 1600 as it is a 7g-grandmother who was of the Blackfoot Nation. Uniparental DNA--yDNA and mtDNA--have their uses in looking for NA origins, but that's a whole 'nuther discussion.
