In genetic association studies we explain the association of a polymorphic allele or genotype in terms of odds ratio i.e., the odds of getting a disease when compared with a reference allele or genotype. Usually we take the ancestral allele (also called as wild allele) as reference and it also used to be the major allele (i.e higher in frequency). There could be situations where the ancestral allele is not the major allele in the population. In that case which one should I choose as reference allele? Is the major allele irrespective of ancestral nature or the ancestral allele irrespective of the frequency ?
The reference allele is by definition the allele present on the plus strand of the human reference genome sequence. The ancestral allele, on the other hand, is determined as the allele present in the last common ancestor of humans and great apes. Therefore, the major allele in a human population is not necessarily the ancestral allele and vice versa. AFAIK, there is no gold standard for which allele to report in a GWAS. Frequently, the minor allele in the study cohort is reported as the alternative (ie non-reference) or effect allele. In general, the selection of the effect allele is not problematic as long as you always refer to effect estimates and effect directions as being allele-specific and also mention the respective allele, as then your results are unambiguous. (e.g.: per copy of the A allele there is an x-fold increase of risk ...).
The reference allele is by definition the allele present on the plus strand of the human reference genome sequence. The ancestral allele, on the other hand, is determined as the allele present in the last common ancestor of humans and great apes. Therefore, the major allele in a human population is not necessarily the ancestral allele and vice versa. AFAIK, there is no gold standard for which allele to report in a GWAS. Frequently, the minor allele in the study cohort is reported as the alternative (ie non-reference) or effect allele. In general, the selection of the effect allele is not problematic as long as you always refer to effect estimates and effect directions as being allele-specific and also mention the respective allele, as then your results are unambiguous. (e.g.: per copy of the A allele there is an x-fold increase of risk ...).
I don't think it matters, as long as the allele is unambiguous, and the reference and effect allele are stated. I usually avoid complementary mutations (A/T, or C/G), but if I were to use them I would include surrounding bases to ensure that the strand could be inferred from the sequence. I'm not a big fan of choosing "human genome plus strand", because that can change for different builds. If you do that, then you need to also state the build number that you're using for reference.
- Badges
- Science topic
- Similar topics
- Biological Science
- Genetics
- Genetic Analysis