I am new in population genetics. I am working on a coding region of Human gene. I have found 5 human specific substitutions out of these 1 substitution is polymorphic in human population.
I wanted to apply selection test (ihs/ehh) on this SNP. The frequency of ancestral and derived alleles for this particular snp is as follow:
All pop: Ancestral Allele:2% Derived Allele:98%
Africa: Ancestral Allele:8% Derived Allele:92%
America: Ancestral Allele:0% Derived Allele:100%
Asia: Ancestral Allele:0% Derived Allele:100%
Europe: Ancestral Allele:0% Derived Allele:100%
Can I apply ihs/ehh only on African population
Kindly help me I will be thankful.
The EHH/iHS is a method typically applied within populations (unlike Fst which always requires a comparison between populations). However, you'll need more than one SNP to reconstruct the haplotype (EHH: extended haplotype homozygosity) around the focal mutation, and then compare the homozygosity in this region to the overall homozygosity along the rest of the chromosome/genome (iHS: integrated haplotype score).
This, to put it very simply in a few words: more details can be found in the papers by Sabeti, Voight and several other authors who worked on this topic.
Hope this helps!
Filippo
Dear Anabia,
Based on the frequency in Africa, you can use iHS, but you will have limited statistical power. Indeed iHS has maximal power when the derived allele frequency is between 50% and 75%. At 92% frequency the statistical power is really not good. What I would recommend is to use the XP-EHH statistic instead, using non-African populations as the control.
However given that the derived allele is at 100% frequency in non-Africans I doubt that you will detect any hint of selection with a haplotype based test. If you had a selective sweep in your case driving the derived allele to fixation in non-African and close to fixation in African populations, it is a selective event that took place too long ago to be detected by haplotype based tests. I would recommend you look at statistics such as Tajima's D or other statistics based on the site frequency spectrum, that are better suited to detect complete, species-wide sweeps. You don't even have to do it yourself, there are databases available online of such statistics calculated from the 1,000 Project data.