Experiences with Family Finder and PF

[Yaffa]

I have heard you have to have the roots to get useful DNA samples. It probably is much more expensive too if you don't use the standard spit or swab samples.

I heard this also.

[Yaffa]

That's not always the case. Sometimes it's possible to get mtDNA samples from hair. Of course, that presumes you have a hair sample, but keeping them was popular in the 19th century.

Well yes if someone has a good DNA sample that might have been left by that deceased male. I dont know any living family of mine who might have DNA samples of hair on deceased relatives. Dont they need the root of the hair not just a clipping of a hair sample??

[smallaxe]

That's not always the case. Sometimes it's possible to get mtDNA samples from hair. Of course, that presumes you have a hair sample, but keeping them was popular in the 19th century.

I have heard you have to have the roots to get useful DNA samples. It probably is much more expensive too if you don't use the standard spit or swab samples.

[smallaxe]

Thanks, I hadn't seen that website before. And keep us posted on your software. (You're not making it Windows specific I hope.)

I'm writing it in C# (.NET), so it should run on linux and MaxOS if you have the Mono framework installed.

[nathanm]

You cant get MT-DNA on deceased men.

That's not always the case. Sometimes it's possible to get mtDNA samples from hair. Of course, that presumes you have a hair sample, but keeping them was popular in the 19th century.

[nathanm]

Here is how I have done it in the past. You open up your raw data file in notepad++ (it's too big for most spreadsheet programs). Find the area using the chromosome and position where the interesting segment is. Select the rows in the segment (plus some before and after the segment) and copy to the clipboard. Now paste that into a spreadsheet, or paste it into a text editor and save it as a .csv file and open that file in a spreadsheet.

So now you have a spreadsheet showing the data for the area of your genome where the segment of interest is. Select the values in the RSID column (the SNP names that start with "rs"), or if there are many of them select up to 1000 of them at a time. Open the following link in a web browser:

http://spsmart.cesga.es/

....

I'm a programmer, and I'm currently working on a program that will do this automatically, and that people can run on their own computers so they don't have to send me their raw data.

Thanks, I hadn't seen that website before. And keep us posted on your software. (You're not making it Windows specific I hope.)

[nathanm]

Also, I have heard that if a scenario exists where the last 100% Native American Indian existed in a person’s genetic line more than 6 generations ago, it is difficult to detect using Population Finder because of the percentage being too low to detect. Maybe PF will develop higher resolution tests in the future that are capable of detecting a low percentage of NA more than 6 generations ago.

Unfortunately, it's not a matter of test resolution. There's no way to increase the resolution, it's already as finely grained as possible, comparing individual base pairs (A, C, G, or T) against known SNPs. The problem is two-fold, and its origin lies elsewhere.

First, there just haven't been enough people tested, from various populations, to be as accurate as any of us would like. There are better reference populations than the HGDP, namely the HapMap and 1000 Genomes projects. Someday soon, I hope, FTDNA will switch to using one or more of those databases for PF, or even a combination of all of them. Also, as more genetic research is done, new SNPs will be identified that are predominantly found in certain populations. Then it'll be a matter of comparing your data with the new SNPs, which were hopefully tested by the Illumina chip.

Second, and more significant, is the simple arithmetic of inheritance, and probability. After six generations, the average contribution of each of those 64 ancestors is only 1/64 of your DNA, or a little over 1.5%. After just 5 generations, the probability you inherited DNA from any individual ancestor is about 50/50. Because of DNA recombination, the amount inherited from each ancestor can be much different than the theoretical average. We should have about 25% from each grandparent, but the actual share might be 20%, or it could be 40%. When you go back a few generations, with average contributions in the single digits, the actual contribution could very well be zero. However, even if you didn't match a known ancestry, one of your siblings might.

[Yaffa]

I had missed that. I made an assumption based on his name. You are correct about determining tribe via DNA. It's hard enough to identify NA vs. East Asian. I don't think any test can identify down to the level of tribe, and if a test claimed it could, I would be highly skeptical. PF is not making that claim. It's only reporting the sample populations a person is most like, and Pima/Maya is about it for North America as far as sample populations go. Y-DNA and MtDNA are good when they are positive for NA haplogroups, but they can't disprove NA ancestry since they represent less than 2% of your ancestry 7 generations back.

I also agree with what I think you are saying, i.e. that claiming Native American ancestors was the safe explanation among families in the south to explain physical traits actually attributable to African ancestors. This is just my opinion based on doing southern genealogical research for many years. I'm not saying this is the case for DelawareNative, I'm speaking generally now.

With Y and MT DNA, you would have to get Y and MT DNA on each or as many direct line ancestors as you can to tell a full story. Lines dies out. You also cant get Y DNA on women. You cant get MT-DNA on deceased men. You never get the full story.

To have an Indian ancestor has to be within 500-600 years, its then finding paper where it falls on the tree.

I can prove my closer relatives great grandparents to me are either Indian or European through Paper and Y and MT DNA. Now going back further to my 5th/ 6th greats or before, my European men do have ties to and I have circumstantial evidence that I might have some free African American in my lines. I have no solid evidence at the moment . These possible free African Americans on my tree going that far back would not show up on a PF test. Im missing marriages. This is where I would need MT-DNA on those missing wives to prove they might have been African.

as for the Pima/Maya. They have that DNA because they were doing a diabetes study. They say the Pima have one of the highest rates of diabetes. I also read they did a DNA study on the early settlers of Starr County Texas where they too have one of the highest rates of diabetes. Those settlers they claim carry the specific Pima/Maya markers. My ancestors being one of those early settlers although they are not Pima/Maya. Many of my ancestors have died due to serious complications of diabetes in the past. That Pima/Maya marker maybe specific to diabetes????? I have heard that many Indian tribes in general came down with diabetes due to changes in their diet when their natural food sources were taken by invasion by the Europeans.

[smallaxe]

Interesting. I'm now in the same position, trying to make sense of an unexpected NA that Dr. McDonald identified as one "small but strong" segment (plus a few other spots that are more likely "noise.")

So how do you go about studying this?

(Sorry if my question is naive

Blair

Here is how I have done it in the past. You open up your raw data file in notepad++ (it's too big for most spreadsheet programs). Find the area using the chromosome and position where the interesting segment is. Select the rows in the segment (plus some before and after the segment) and copy to the clipboard. Now paste that into a spreadsheet, or paste it into a text editor and save it as a .csv file and open that file in a spreadsheet.

So now you have a spreadsheet showing the data for the area of your genome where the segment of interest is. Select the values in the RSID column (the SNP names that start with "rs"), or if there are many of them select up to 1000 of them at a time. Open the following link in a web browser:

http://spsmart.cesga.es/

Check the "CEPH U. Stanford HGDP " box and click the Metasearch button. In the next screen, check the America, Africa, and all the other continental area boxes. In the "Search by SNPs" box, paste the RSID values you copied from your spreadsheet, and then click the "next" button. On the next page, just click the "search" button. After this you will see a page with all the SNPs you pasted, along with pie charts that show the continental frequencies of the alleles. You can look at your values for each SNP in your spreadsheet (remember that A=T and C=G). If there is an SNP that is a good indicator for a continent of interest, and you have the value that is most likely associated with that value, then you can click on the rs name on the left hand side. This will open a web page that tells you everything about that SNP including a more detailed breakdown of population frequencies among the sub-populations, and also other sets of population samples if there are any.

If you are looking mostly European looking for non-Euro admixture, you can sort your spreadsheet first by alleles (AA, AC, CC, TC, etc), secondly by chromosome, and thirdly by position. Then select a block of heterozygous SNPs (alleles like AC were the two values are different, but not like CC where the two values are the same). These are the most likely SNPs to show admixture. This is because if you are homozygous, then either you're 100% a particular population, or you are 50/50 two different populations, but both alleles are relatively frequent for that SNP. But the heterozygous ones are most likely where you'll find that one of the alleles is infrequent in one population, but frequent in the other, and since you're heterozygous, it supports that you're 50/50 for two different populations. Also note that there are contra-indicators too. For example, if you are homozygous CC at some SNP, but 100% of Africans are AA, then this contraindicates you being part African there. Multiple contra-indicators within a potential segment can put the whole segment into question. Contra-indicators can also be helpful in determining where a real segment starts and ends. Remember that these are all just statistics. Even if it says 0% of Europeans have a particular allele at an SNP, it may actually be 0.2%. That's 2 out of 1000, and you could be one of them. So one SNP is not enough. You need to see a pattern of supporting SNPs.

I've attached a screen shot of an area where it appears I have a African segment among my mostly European DNA, and another where there is a possible NA segment to show what I mean. I'm heterozygous for all the displayed SNPs. It was hard to find a good screenshot of an NA example, because there are so few good markers, and I'm not sure my segment is really even NA. And the of the ones that show a difference between Euro and NA, there is usually a fair amount of East Asian too, so you can't be sure whether you're seeing actual NA, or Asian bits that sometimes show naturally in Europeans. I added green arrows to that one to point out some possible NA SNPs.

It's not enough to just find one or two possible SNPs. You need to see that they're consistently showing up throughout the segment, and it has to run over a big enough area (>1 mega base pairs) to be more than noise. I'm a programmer, and I'm currently working on a program that will do this automatically, and that people can run on their own computers so they don't have to send me their raw data.

[smallaxe]

Per Delawarenative they were hoping that PF would pin the tribe their ancestors are from since their are no records stating tribe on their ancestors. However no DNA test can prove tribe. Free African Americans also claimed Indian because it was safer. They state their records are stating Negro and white but other researchers can prove the line is Indian (no tribe). Being that there seems to be no solid proof their ancestors are Indian, how do you claim your ancestors are Indian? This is where Y and MT DNA come in handy where DNA would be solid proof of having a possible Indian ancestor.. PF has me classified as Pima/Maya but my ancestor is not of either tribe. So PF is not even pinning my ancestors tribe with accuracy. I can also tell you that my MT-DNA is matching to other tribes that are not my ancestors tribe so again DNA does not prove tribe, only paper does.

I had missed that. I made an assumption based on his name. You are correct about determining tribe via DNA. It's hard enough to identify NA vs. East Asian. I don't think any test can identify down to the level of tribe, and if a test claimed it could, I would be highly skeptical. PF is not making that claim. It's only reporting the sample populations a person is most like, and Pima/Maya is about it for North America as far as sample populations go. Y-DNA and MtDNA are good when they are positive for NA haplogroups, but they can't disprove NA ancestry since they represent less than 2% of your ancestry 7 generations back.

I also agree with what I think you are saying, i.e. that claiming Native American ancestors was the safe explanation among families in the south to explain physical traits actually attributable to African ancestors. This is just my opinion based on doing southern genealogical research for many years. I'm not saying this is the case for DelawareNative, I'm speaking generally now.

[bkilpatrick]

DelawareNative


Related to the above, I'm currently studying some areas of my Aunt's DNA that Dr. McDonald identified as Middle Eastern (we have no known Middle Eastern ancestry, or even any southern European ancestry). Interestingly, one of the MidEast segments is exactly adjacent to a small NA segment. By examining the actual SNPs and their allele population frequencies in that are of the chromosome, I want to see if it's possible that NA DNA might be getting attributed to other populations (or vice versa).

Interesting. I'm now in the same position, trying to make sense of an unexpected NA that Dr. McDonald identified as one "small but strong" segment (plus a few other spots that are more likely "noise.")

So how do you go about studying this?

(Sorry if my question is naive

Blair

[bkilpatrick]

Thank you, Daniel72, for explaining why the PF results are so rough.

Thank you, bkilpatrick, for Dr. McDonald's contact info for requesting participation in the BGA Project:



I politely asked the same of Dr. McDonald and he has obliged me as well. I am looking forward to his analysis!

Thank you all!

You are welcome, and good luck! Let us know what you learn. (Dr. McDonald seems to work quickly

Blair

[Yaffa]

DelawareNative
I don't need to tell you Delaware tribal history, but others may not be aware that the Delawares have been in contact with whites for a long time. In fact, it's a testament to the Delaware people that they have been able to retain a distinct identity after so many years of turmoil and displacement. If you read the first hand accounts of early Rocky Mountain trappers, you quickly pick up on the close association they had with the Delaware, who often made up part of the trapping parties, and were held in extremely high regard by the white trappers.

But this this long interaction, and sometimes close association with whites provides ample opportunity for long term admixture occurring. And the mechanism of DNA transmission from generation to generation means that the DNA of some ancestors beyond 4 generations may not be present or at a detectable quantity, and beyond 5 generations there are many ancestors that are not present genetically. So it may be partly attributable to the roll of the dice that no NA DNA is being picked up for you.

Another problem is that there are not as many SNPs that are clearly distinguishable as NA by the frequency of allele occurrences. It is harder to identify NA segments in a map of a person's chromosomes. Sub Saharan African DNA by comparison, has many more SNPs whose allele frequencies are distinctly African. This is exacerbated by the lack of good, un-admixed samples from which to derive NA allele frequencies. So I think it is possible for someone to have some NA DNA that is undetectable because it happens to be in areas of the DNA where there are insufficient SNP markers suitable for identifying it as NA DNA.

Related to the above, I'm currently studying some areas of my Aunt's DNA that Dr. McDonald identified as Middle Eastern (we have no known Middle Eastern ancestry, or even any southern European ancestry). Interestingly, one of the MidEast segments is exactly adjacent to a small NA segment. By examining the actual SNPs and their allele population frequencies in that are of the chromosome, I want to see if it's possible that NA DNA might be getting attributed to other populations (or vice versa).

The final possibility is that your test results were contaminated or swapped inadvertently. Mistakes do happen. Have you looked into retesting? Or have you been able to corroborate that the results are likely yours because of matches to people confirmed to be related to you on paper?

Per Delawarenative they were hoping that PF would pin the tribe their ancestors are from since their are no records stating tribe on their ancestors. However no DNA test can prove tribe. Free African Americans also claimed Indian because it was safer. They state their records are stating Negro and white but other researchers can prove the line is Indian (no tribe). Being that there seems to be no solid proof their ancestors are Indian, how do you claim your ancestors are Indian? This is where Y and MT DNA come in handy where DNA would be solid proof of having a possible Indian ancestor.. PF has me classified as Pima/Maya but my ancestor is not of either tribe. So PF is not even pinning my ancestors tribe with accuracy. I can also tell you that my MT-DNA is matching to other tribes that are not my ancestors tribe so again DNA does not prove tribe, only paper does.

[smallaxe]

DelawareNative
I don't need to tell you Delaware tribal history, but others may not be aware that the Delawares have been in contact with whites for a long time. In fact, it's a testament to the Delaware people that they have been able to retain a distinct identity after so many years of turmoil and displacement. If you read the first hand accounts of early Rocky Mountain trappers, you quickly pick up on the close association they had with the Delaware, who often made up part of the trapping parties, and were held in extremely high regard by the white trappers.

But this this long interaction, and sometimes close association with whites provides ample opportunity for long term admixture occurring. And the mechanism of DNA transmission from generation to generation means that the DNA of some ancestors beyond 4 generations may not be present or at a detectable quantity, and beyond 5 generations there are many ancestors that are not present genetically. So it may be partly attributable to the roll of the dice that no NA DNA is being picked up for you.

Another problem is that there are not as many SNPs that are clearly distinguishable as NA by the frequency of allele occurrences. It is harder to identify NA segments in a map of a person's chromosomes. Sub Saharan African DNA by comparison, has many more SNPs whose allele frequencies are distinctly African. This is exacerbated by the lack of good, un-admixed samples from which to derive NA allele frequencies. So I think it is possible for someone to have some NA DNA that is undetectable because it happens to be in areas of the DNA where there are insufficient SNP markers suitable for identifying it as NA DNA.

Related to the above, I'm currently studying some areas of my Aunt's DNA that Dr. McDonald identified as Middle Eastern (we have no known Middle Eastern ancestry, or even any southern European ancestry). Interestingly, one of the MidEast segments is exactly adjacent to a small NA segment. By examining the actual SNPs and their allele population frequencies in that are of the chromosome, I want to see if it's possible that NA DNA might be getting attributed to other populations (or vice versa).

The final possibility is that your test results were contaminated or swapped inadvertently. Mistakes do happen. Have you looked into retesting? Or have you been able to corroborate that the results are likely yours because of matches to people confirmed to be related to you on paper?

[Darkwriter]

Thank you...

Thank you, Daniel72, for explaining why the PF results are so rough.

Thank you, bkilpatrick, for Dr. McDonald's contact info for requesting participation in the BGA Project:

In any case, I sent him an e-mail to ask about participating and he kindly agreed.

I politely asked the same of Dr. McDonald and he has obliged me as well. I am looking forward to his analysis!

Thank you all!