1
$\begingroup$

I am new to Admixture and I would like to use fastNGSadmix to look for admixture in one individual (modern). Admixture is supposed to look for proportions of ancestral populations in a sample (either a population or an individual sample). for that I will need to prepare my reference population set to compare with my actual sample which is an individual form region X for instance.

Now I am thinking whether I have to add the X population to my reference set or not? If yes, wouldn't it introduce bias? Like my individual will always show the X population as the first admixing population?

$\endgroup$
2
  • 1
    $\begingroup$ Hello, welcome to Bioinformatics.SE. I have never used Admixture personally, but as far as I know it's supposed to quantify contribution of different populations to genomes (snps) of samples. So it depends on your question. You could maybe add bit more details. We like details. Details make questions answerable. $\endgroup$
    – Kamil S Jaron
    Jan 22 '18 at 16:16
  • $\begingroup$ Hi, well I guess you are right about more detail. Admixture is supposed to look for proportions of ancestral populations in a sample (either a population or an individual sample). for that I will need to prepare my reference population set to compare with my actual sample which is an individual form region X for instance. Now I am thinking whether I have t add the X population to my reference set or not? if yes, wouldn't it introduce bias? like my individual will always show the X population as the first admixing population? I hope it is clear enough :) $\endgroup$
    – FBR
    Jan 23 '18 at 5:54
2
$\begingroup$

Disclaimer, I am not an expert on admixture, answer is provided without any warranty.

It depends what you want to figure out. Why do you want to categorize the ancestry of the individual? If you analyze this origin of population X you should make a reference without population X and analyze all the individuals from the population with all the surrounding populations in the reference. However if you are interested in origin of a single individual, than the population X got to be in reference as well.

If your result will be that 90% of SNPs of your individual correspond to your population X, then it means that's a normal X fellow. Why it should be a problem?

-- edit --

I just found this preprint that highlights potential pitfalls in interpretation of STRUCTURE/ADMIXTURE results.

$\endgroup$
1
  • $\begingroup$ Thank you. well, my goal is to determine the proportions of ancestral populations and my problem with the 90% match to the source population is that I am not really sure if this 90% means that the source population in ancestral to the individual. does it? It is basically about similarities. If I compare sth with itself I will always get the best match. no? $\endgroup$
    – FBR
    Jan 23 '18 at 12:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.