DNAprint Genomics (DNAG)

[mingwan0]

BGA groupings in ANCESTRYbyDNA 3.0

This is a complicated topic, mainly due to the fact that there seem to be several taxonomic classifications for human "races". The current product is limited to four categories, Sub-Saharan African, Native American, East Asian, and Indo-European. What categories will be in future versions? Some relevant reference material is copied below. My further comments are in italics.

http://www.ancestrybydna.com/

Ongoing research at DNAPrint has suggested a more logical grouping based on the four BGA delineations shown on the map in the home page – where South Asian, Middle Eastern and European are grouped into a single group called Indo – European. This grouping makes sense in light of anthropological evidence and cultural connections between these groups (for example, their languages are derived from a common base), and our data suggest these groups are far more similar to one another in genetic sequence content than to other groups. We have also found that the test performs more accurately when Pacific Islanders are grouped with East Asians. Because our desire is to produce the most accurate results possible, we will use the following four groupings for the ANCESTRYbyDNA 2.0 test: 1) Native American (those peoples that migrated to inhabit South and North America), 2) Indo – European (Europeans, Middle Easterners and South Asians such as Indians, 3) East Asians (Japanese, Chinese, Koreans, Pacific Islanders) and 4) Africans (sub-saharan). Constructive criticism from other Molecular Anthropologists have also suggested this new 4-way grouping is a more reasonable task for our current test, and we agree. Thus, we have decided to wait to attempt to distinguish between South Asian and European, as well as Pacific Islander from East Asian with the ANCESTRYbyDNA 3.0 test.

http://www.familytreemagazine.com/articles/aug03/dna.html

A branch of Frudakis' company called AncestryByDNA aims to help genealogists explore the racial mix of their family trees with a simple DNA test. Unlike other DNA-testing kits, AncestryByDNA does not rely on Y-chromosome tests (which only males can use) or mitochondrial DNA. Instead, it looks at a person's Single Nucleotide Polymorphisms (SNPs, or "snips" for short)—think of them as collections of letters among the long sentences of the human genome. Then, AncestryByDNA compares your SNPs to a database of results representing four main human racial groups, based on continent of origin: sub-Saharan African, Indo-European (Europeans, Middle Easterners and South Asians, such as Indians), East Asian (Japanese, Chinese, Koreans, Pacific Islanders) and Native American (ancient migrants to both North and South America). Originally, the test used six groups, distinguishing South Asian from European and Pacific Islander from East Asian, but the current version retreats from that until further refinement in an upcoming version 3.0.

http://archiver.rootsweb.com/th/read/GENEALOGY-DNA/2003-08/1061655028

It's my understanding that the next test will be called 2.5, not 3.0.

The 2.5 version will add a number of markers, but it will still be using the same four population divisions (Indo-European, Native American, African, and East Asian). The additional markers will tighten the confidence intervals. We'll undoubtedly be discussing whether an upgrade is worthwhile for Ancestry 2.0 customers.

The 3.0 version will subdivide those four categories into more groups. They originally planned to have six, including Pacific Islander and South Asian, as I recall, but decided that version 2.0 didn't have that kind of resolution. I don't know what the timetable is for 3.0

So, it looks as if a six-way grouping would be Sub-Saharan African, Native American, East Asian, European, South Asian and Pacific Islander. If this was to be expanded it looks as if the following nine groups might be implied:

Sub-Saharan African
Europeans, Middle Easterner, South Asians
Japanese, Chinese, Korean, Pacific Islander
Native American

http://www.dnaprint.com/pr_9_19.html

DNAPrint scientists are collaborating with Dr. Shriver to develop more advanced versions of ANCESTRYbyDNA that may be useful for discerning regional heritage proportions in individuals. For example, ANCESTRYbyDNA 3.0 is expected to be capable in the near future of determining whether an individual is of Irish/British, Middle European (French, German), Scandinavian, Mediterranean (Italian, Greek, Spanish) or Eastern European heritage as well as of Western/Central versus East African heritage or of Japanese, Chinese or Korean heritage. For more information about the ANCESTRYbyDNA 2.0 service please visit http://www.ancestrybydna.com.

So now we have an implied fourteen-way grouping:

Western/Central African, East African
Irish/British, Middle European, Scandinavian, Mediterranean, East European
Middle Easterner, South Asians
Japanese, Chinese, Korean, Pacific Islander
Native American

Which would become a seventeen-way grouping if they were able to sub-classify the Middle European and Mediterranean groups.

http://www.dnaprint.com/2003/corporate/participate.html

We are working to expand our current base of four (4) distinctions to twenty (20).

Close enough perhaps to suggest the sort of groupings that DNAP might have in mind for version 3.0? I thought so until I saw the following.

http://archiver.rootsweb.com/th/read/Melungeon/2003-10/1065754240

The recent questions here about the so-called DNA print test prompted some back-channel discussion in search of authoritative answers, which in turn prompted Mike Nassau, a population geneticist himself, to research this topic. His rather lengthy response is below. I certainly want to thank Mike for all his work on this and Karlton Douglas for his assistance as well. The bottom line is, if you thinking of having this test done, wait for version 3.0 at least, for reasons Mike explains in detail.

Dennis Maggard
List Owner

October 9, 2003

Dear Friends,

I got an e-mail today from Karlton Douglas asking my opinion about the new ANCESTRYbyDNA test offered by DNAPrint Genomics. Being me, I wrote a three page response to his one line query. I thought some of the rest of you might be interested. Since writing, I have read that they will soon bring out ANCESTRYbyDNA3, which will be able to distinguish Mediterranean from Northwestern European. If you are interested in doing this test, I would strongly recommend that you wait until version 3 is ready. Maybe it will also be able to distinguish at least Na-Dene from Amerind types of Native Americans, which would greatly increase my confidence that it can distinguish Native American from North and East Asian. My earlier paper on Race and Miscegenation, written a year ago for a discussion on the Melungeon Family Genealogy Forum about the validity of race, is now attached as a file to this group, as is the write-up where I learned about the new version 3 of the test.

Any of you may feel free to copy this or part of it to any other list, but answers or questions should be directed back to the Melungeon Origin group.

http://groups.yahoo.com/group/MelungeonOrigin

[e-mail]

Mike Nassau

P.S. October 10,2003

I have had an interesting response indicating that version 2.0 can not distinguish Amerind from "East Asian" , that someone of known Algonquian ancestry was told they were East Asian, not Native American. This is exactly what I would expect from a test which does not distinguish Na-Dene (Navajo, Apache, Athbaskan, Tlingit and other groups which came to America from Asia only some 8000 years ago) from Amerind (the rest of Native Americans, whose ancestors came over twice as long ago). These two groups are closer to related groups in North and East Asia than they are to each other.

I now have had time to wade through the blurbs put out by DNAPrint including the one on their "science". Their "science" article is reassuringly sound and well written. They do know what they are doing. Their work will tell people a lot about their ancestral composition, with one major problem or exception. First, wait for version 3.0, version 2.0 doesn't distinguish between different "Indo-European" groups. (They seem to lump all "white" or "Caucasian" racial groups under this umbrella, which is very wrong.) If version 3.0 will distinguish Na-Dene from Amerind, and both of them from related groups in North Asia, it will serve for people of mixed NW European, Mediterranean, Sub-Saharan Black and Native American ancestry given that the mixing has not been followed by visual selection. That proviso pretty well rules it out for Mestees old mixed groups) like Melungeons [including Carmel Indians and Goinstown Indians], Brass Ankles, Redbones, etc., but maybe not for Lumbees, Haliwa Indians, Occaneechi Saponi, etc. The difference is that the lighter groups like the Melungeons and Brass Ankles have been visually selected, with the more "Black" appearing members tending to leave and join Blcak communities. This means that the appearance has become more "White" over time while the actual ancestry hasn't changed.

Pollitzer's blood type studies had the advantage that blood types are not seen and people are not selected based on their blood type. So visual selection can explain why a group like the Hancock County Melungeons he tested were over one eighth Black in ancestry but appeared much less. This study, with its use of Duffy blood type, which is found only in people of Black West African ancestry, is still the definitive study on Melungeon ancestry. Guthrie's reanalysis in no way diminishes this finding of a large Black component, what it does show is that if Pollitzer had included a Mediterranean component like Portuguese in his model instead of just Black, English and Cherokee, he would have had a much better fit. That better fit would have been at the expense of the English, not the Black component.

ANCESTRYbyDNA uses some 2000 SNPs (Single Nucleotide Polymorphisms) to type people for their ancestry. The problem is that they have not used only sites which do not impact appearance. In fact, they have looked for sites which do have an effect on appearance. So visual selection will shift the results of the test significantly. So a Melungeon who is 14% Black, 1% Native American, 25% Mediterranean and 60% NW European in ancestry may well test out to be 5% Black, 0% Native American, 30% Mediterranean and 65% NW European.

Be warned. The advertisements for this product are dishonest and deceptive. Do not believe what you read. I think this more likely the fault of the advertising agency than of the scientists, but it is very real. When they say this test will tell you your precise ancestral mixture, that is patently false. Any scientist or statistician would be looking for their statistical margin of error. My guess is with 2000 SNPs used, they would have about 1 to 3% error in their estimate for each generation since the mixing occurred. So for people whose ancestry was recently mixed, the estimates will be pretty good, for those with old mixing, not so good.

I do hope this test is improved in future versions. I particularly hope they may bring out a version which excludes SNPs in loci for visually identifiable traits.

Mike

Dear Karlton,

I have just started to look at the info I have collected on the internet. The genetics, using a wide variety of SNPs (Single Nucleotide Polymorphisms), seems valid enough. What I have read of their interpretation of the data seems very, very simplistic at best, racist and ignorant at worst.

The immediate glaring problem is their lumping of people into races or groups. South Chinese are closer to Pacific Islanders than they are to North Chinese. North Chinese are closer to Navajo than they are to Japanese. Navajo are closer to Chinese than they are their Hopi neighbors. Eskaleuts (inuit, Yupik, Aleuts) are very close indeed to Chukchi, Kamchatkans and other peoples of extreme NE Asia and closer to Uralic peoples like Finns and Samoyeds and Indo-Europeans like Russians and English than they are to Amerinds like the Algonquians and Siouans. These Amerinds are probably closer to Kets and Huns (language extinct, but many descendants) than to Na-Dene (Athbaskans, Apache, Navajo).

The races of humankind are divided first by the direction from the point of origin of modern man, somewhere around Lake Victoria, probably in Kenya or Uganda. To the south are the two Southern African races, the Twa (pygmies) and Khoisan (San ["Bushmen"] and Khoi-Khoi ["Hottentots"]). To the west and northwest are the two Black African races, Guinean and Sahelian. To the east and northeast are the Trans-Saharan races. The Trans-Saharan can be broken into the Australoid, the Papuan [New Guinea and related areas], the Austric (Southeast Asia, South China and the Austronesian [Malayo-Polynesian] area) and the Rest. The Rest includes North Africa and the Horn of Africa, Eurasia except for SE Asia and South China, and the Americas. Coming up with a name which makes sense for a group as diverse as this, including Icelanders, Basques, Moors, Somalis, Dravidians, Kets, Finns, North Chinese, Turks, Japanese, Inuit, Navajo and Quechua, is very difficult.

I do not see these eight races in AncestorsbyDNA, rather, I see the traditional breakdown slightly modified. I am particularly disturbed by the treatment of Native Americans. If they have SNPs which will lump Na-Dene with Amerinds instead of North Chinese, I would be very surprised. The Far East race is another puzzle. If this is meant to include Japanese (with their affinities to Turkics and Eskaleut), the North Chinese and the South Chinese, I do not believe they have the common ancestry to provide SNP common identity.

OK. So this implies an eight-way grouping:

Twa, Khoisan
Guinean, Sahelian
Australoind, Papuan, Austric, "the Rest"

Other sub-divisions, such as distinguishing between Na-Dene and Amerind (other Native Americans) also seem to be important. If we take the groupings from the last paragraph above we would have the following twelve-way grouping (ten-way if the Austric group is not sub-divided):

Twa, Khoisan
Guinean, Sahelian
Australoind, Papuan
Southeast Asia, South China, Austronesian
North Africa/Horn, Eurasia, Americas

However, the implied further sub-divisions within "the Rest", some of which match the categories mentioned by DNAP, would expand this to at least twenty-two groups ("The Rest includes..."):

Twa, Khoisan
Guinean, Sahelian
Australoind, Papuan
Southeast Asia, South China, Austronesian
Icelanders, Basques, Moors, Somalis, Dravidians, Kets, Finns,
North Chinese, Turks, Japanese, Inuit, Navajo, Quechua

Again, if the Austric group is not sub-divided this becomes a twenty-way grouping...

I think this is coincidence and not necessarily the same twenty groups that DNAP has in mind. It will though be interesting to see what they eventually are. Finally, a more definitive fifty-two-way classification:

http://www.legioneuropa.org/mxrabd/rosenberg.pdf

Genetic structure of human populations", Rosenberg et al., Science 298, pgs. 2381-2385

African: Bantu, Mandenka, Yoruba, San, Mbuti Pygmy, Biaka Pygmy

Europe: Orcadian, Adygei, Russian, Basque, French, Italian, Sardinian, Tuscan

Middle East: Mozabite, Bedouin, Druze, Palestinian

Central/South Asia: Balochi, Brahui, Makrani, Sindhi, Pathan, Burasho, Hazara, Uygur, Kalash

East Asia: Han, Han (N. China), Daj, Daur, Hezhen, Lahu, Miao, Oroqen, She, Tujia, Tu, Xibo, Yi, Mongola, Naxi, Cambodian, Japanese, Yakut

Oceania: Melanesian, Papuan

America: Karitiana, Surui, Colombian, Maya, Pima