ADMIXTURE results - how can I determine what SNPs are associated with different inferred ancestry components?

Question

[devenvyas]

I have some ADMIXTURE results that I am struggling to plot, but I am interested in determining what SNPs are going into which components. I know that ADMIXTURE itself does not specify which components correspond to which SNPs. I was wondering if anyone has any suggestions on how I could do this.

I am looking into ELAI, LAMP, and HAPMIX, but I am open to suggestions.

Thanks!

Answer 1

[Gabriel R.]

You don't really have SNPs associated with a particular ancestral component but rather, an estimate of allele frequency in each ancestral component. That's the P file. See the manual page 3:

https://www.genetics.ucla.edu/software/admixture/admixture-manual.pdf

[devenvyas]

I already know that, I want to know if there is an alternate program I can use to cut portions of the genome into different ancestries.

[Gabriel R.]

Ah then have a look at ChromoPainter and fineStructure

http://www.paintmychromosomes.com/

[LauferVA]

In that case, many of the programs you listed are a really good place to start, essentially you are asking about local ancestry estimation, for which ELAI, LAMP, HAPMIX, chromosome painter etc. are good alternatives.

There also programs that use local ancestry to increase power for association testing if that is of interest to you, in this case maybe start by reading http://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1002325

The same author wrote a fairly good review in 2013 .. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3556814/

[devenvyas]

I'll look into chromopainter as well then. I am not really interested in association testing. I am just looking for Neanderthal in different parts of the genome.

[Gabriel R.]

I remember Sriram Sankararaman did something along those lines for the Altai Neanderthal paper. Have a look at this maybe:

http://www.nature.com/nature/journal/v505/n7481/full/nature12886.html#/denisovan-gene-flow-in-mainland-asia

He essentially used an HMM to identify these chunks in phased genomes.