Admixtures

What do you think about ADMIXTURE?

  • I can make it work, just let me have a go!

    Votes: 0 0.0%

  • Total voters
    22

Dorianfinder

Regular Member
Messages
472
Reaction score
49
Points
28
Location
Western Cape
Ethnic group
European
Y-DNA haplogroup
R-FGC13617
mtDNA haplogroup
T2b1
Latest version of Admixture http://www.genetics.ucla.edu/software/admixture/download.html

ADMIXTURE is a computer software program that once calibrated by a knowledgeable tutor, is used to estimate an individual's ancestral composition by analyzing the allele frequencies from autosomal SNP data.

The ADMIXTURE program is used to estimate the number of underlying populations through a process of cross-correlation experiments. To do this, individuals of known ancestry are exploited to yield more precise ancestry estimates.

Once ancestral components have been uncovered, small admixture coefficients are eradicated for each individual, encouraging model parsimony and often yielding more interpretable results for small data-sets or data-sets with dissimilar populations.
BMC Bioinformatics 2011, 12:246doi:10.1186/1471-2105-12-246

ADMIXTURE is currently the platform used by most hobbyists and amateur population geneticists.
 
Here are the free tools available for calculating admixture percentages using 23andMe, FTDNA and deCODEme genotype data.

Using the Do-It-Yourself Dodecad v 2.1 that you can find here [includes the Dodecad v3 calculator files]

You are able to calculate your autosomal admixture percentages using the following calculators:

Download the Dodecad v3 calculator here
Download the Euro7 calculator by Dodecad here
Download the BAT calculator by Dodecad here
Download the weac calculator by Dodecad here
Download the Africa9 calculator by Dodecad here
Download the Eurasia7 calculator by Dodecad here
Download the Euro-DNA-Calc111 calculator by Dodecad here
Download the Eurogenes Eurasian K=10 by Davidski here
Download the Eurogenes K=14 by Davidski here
Download the MDLP K=7 by Vadim Verenich here

Download the Dodecad Oracle v1 calculator here

The Dodecad Oracle v1 tool allows you to view your closest population matches and for those with unknown mixed ancestry it lists mixed population matches.

Dodecad software is provided free of charge for non-commercial use by Dienekes Pontikos @ http://dodecad.blogspot.com

If you are having trouble using the program I would be happy to help if I can.
 
Last edited:
You should add the recent Eurasia7, if you check the post you'll see there's also a Calculator for it (several individuals were not included).

Also, there was a Calculator for the Eurogenes K=14, but Davidski removed the post. I don't know if the link is available somewhere, but I have the ZIP. If someone wants to calculate it, I could upload the file. I think it's better than the K=10 in some aspects.

Another interesting point, is that I recently wrote to Dienekes' proposing him to provide tools for the calculation of samples from data source (HGDP, Behar, etc.) using DIY. That will allow people to compare personal results, not only with known people joining the projects, also with those anonimous individuals whose results are only visible sometimes in the population portraits (Dodecad v3) and the Eurogenes project. Just note in the Euro7, Eurasia7, Weac, etc., they are listed as average, but not individually. Then, if Dienekes' finally finds the way to convert this files into a compatible format for DIY, we could see exactly their admixture proportions in the different tests.

For the moment he is busy and told me it can't be done easily, but I'm sure he'll try to do so :)
 
The Eurasia7 by Dodecad and the Eurogenes K=14 by Davidski have been added to the list.
 
Last edited:
I uploaded the K=14 here: https://docs.google.com/open?id=0B8-Sh-LdY4_HYTkxYTAxMzEtNzRlOS00ODNmLWE5ZGQtMDJmM2Y5YTNiNzA0

You can list the link together with the others, Dorian.

The components are:
- Northeast Euro (peaks in Russians)
- West Euro (Basques & Irish)
- East Mediterranean (Cypriots)
- Volga-Ural (Chuvashs)
- West Asian (Lezgins)
- Middle Eastern (Saudis)
- Berber (Mozabite Berbers)
- South Asian (South Indians)
- North Asian (Nganassans)
- South Siberian (Evenks)
- East Siberian (Koryaks)
- East Asian (Japanese)
- Southeast Asian (Southern Chinese)
- Sub-Saharan (Western Africans)

Enjoy!

UPDATE: To run the K=14 the word "test" instead "dv3" is required. That's all.
 
Last edited:
I also uploaded the last version of the Euro-DNA-Calc here: https://docs.google.com/open?id=0B8-Sh-LdY4_HYjc5ODZiMjktOGExYy00ZWQyLWE0YTItMWZjZjg0NjY1ZTBl

It can be added at the beggining too :)

Components:
- Northwest Euro
- Southeast Euro
- Ashkenazi Jewish

It's very simple and must be taken with caution, since sometimes the reports don't make any sense. Keep in mind that Northwest Euro can include plain Western European allele frequencies (I know that because I got 75% there being Iberian), and having some West Asian admixture could report Ashkenazi Jewish here. Just this, be carefull with the interpretations, and pay attention to the major percents.
 
Thanks! The Euro-DNA-Calc software is compatible with 23andMe and deCODEme genotype data whereas the other calculators found in the list in post #2 are all compatible with FTDNA, 23andMe and deCODEme genotype data.
 
Last edited:
ADVANCED ANALYSIS

All 22 chromosomes can be analyzed using the 'bychr' mode instead of the 'genomewide' mode in the dv3.par file found in all dodecad calculators mentioned above. Similarly, ancestry percentages for individual segments of an individual's genome can be calculated using the 'byseg' mode.

The accuracy of analysis increases when viewing all 22 chromosomes. Also, one can study the individual chromosomes and create a graphic representation of each chromosome's ancestral admixture.

If you have used this software please feel free to post your experience here.
 
Last edited:
SNP Map

SNP Map

This tool allows you to browse your individual snp frequencies, and allows you to compare your particular snp to its frequency within other population groups.

The program is downloaded onto your pc and you upload your raw data files from either Relative Finder or Family Finder onto the program. It's user-interface is a no-frills database but it does the job well. It allows you to analyze your individual snps with full confidentiality and it's free.
 
I already checked it quickly with different raw data's from people I know: 1 mixed Iberian (different regions), 1 mostly Catalan (me) a Basque individual and an Angloamerican of mostly English/Irish descent.

I looked at Chromosome X in all cases to see it under similar conditions, and it basically shows there's apreciable shared ancestry between populations. Not all markers are informative in ancestral terms, some of them are related to health or something like this, but I'm not a professional to distinguish them. However, I did not see huge differences between this 4 individuals, it's not rare for example to find a genotype present in 90% of African population and, lets say, only 20% in Europe. At least, I could apreciate it in all the individuals, but perhaps the "sample size" it's too low.

Well, there are a lot of markers one could check with different interpretations. There's nothing more I could say considering my knowledge, just some observations. Other people much versed in allele frequencies would be of greater help.
 
I did not see huge differences between this 4 individuals, it's not rare for example to find a genotype present in 90% of African population and, lets say, only 20% in Europe. At least, I could apreciate it in all the individuals, but perhaps the "sample size" it's too low.

The 20% found in Europe and Africa refers to shared admixture (convergence). If you want to decrease this overlap you can increase reliability from 80% to 95%. The Relative Finder X-chromosome data file has many non-ancestral markers, unlike the FTDNA Family Finder data file which is ancestral in its entirety.

Allele frequencies can be considered reliable when a particular admixture component is found in progressive strands spanning more than 1mil, if it's unbroken over 6mil then ancestry is certain.

Many people don't know what to do with the X-chromosome data file they get from RF and FF. In fact, it's much better than an mtdna test and the best indication of maternal ancestry. The X-chromosome is passed onto men by their mothers and includes all maternal great-grandmothers, so it's amazingly useful for individuals from the USA and South Africa who think they may have a great-grandmother who may have had possible Amerindian or slave ancestry.

X-chromosome pedigree chart.png
 
Thanks. All the raw data's I checked are from 23andme, which of course have many non ancestral markers. Anyways, I'll consider your suggestions and continue exploring similarities.

I think the idea of the program it's quite good, and it could be helpfull for the design of new analysis. Possibly both Dienekes' and Davidski used it for something.
 
New Dodecad Tools

Download the K12a calculator here
Download the world9 calculator here

Download the Dodecad Oracle (K12a version) here (left click 'file' to download)
 
Last edited:
Thanks Dorian, I was about to miss the World9. Interesting this one, I'm going to start a thread for comparison with the previous Eurasia7.
 
Help with the program

Hello,

Could someone please tell me how to use the the Dodecad program for windows, it's says unzip your file to a working directory, I don't get what it is talking about, I just can't seem to get the program to work, could someone give me some instructions that are easy to understand.


Thank you
aaliy06
 
You must create a new folder and call it "working directory", that's what it means. Then, extract the analysis components into this one and continue following the instructions. You need to add the DIY archive and your transformed genotype data (in a format understood by DIY).

Note that previously you must download the latest version of DIY and the R program. I'm assuming you did.
 
Yes I downloaded the latest version, does it matter where you put the working directory folder at? I mean can you put it in your documents or does it have to be on your C: drive? and do you have to have the quotation marks?

Thanks
aaliy06
 
I think it doesn't matter where you create the folder. If it doesn't work, just change the position. Mine are in my documents (I have many working directories for the different calculators) and there's no problem.

The format that DIY understands is the transformation you'll have to do on your raw data following the instructions. Unzip your data inside the working directory, follow the preliminary steps, and an archive called "genotype" will emerge inside the working directory. This one will be valid for all calculators designed for DIY; the only one which DIY is able to read and extract admixture proportions.
 

This thread has been viewed 37023 times.

Back
Top