Non-Additive Associations
September 13, 2017
Citizen Science
September 13, 2017


*Image Credit: Price, A. et. al. Sensitive Detection of Chromosamal Segments of Distinct Ancestry in Admixed Populations. PLoS Genetics 2009

Haplotype-wide Association Studies?

Genome-wide association studies have found 1000s of genetic variants associated with phenotypes of interested. However, the variants found do not explain all of the variance that is known to be due to additive genetics.  One reason for this is that the genetic drivers of risk are not independent single variants but rather regions of the genome where biological function can be impeded in many ways resulting in hard to detect complex genetic architectures.

Rather than try to test all possible combinations of genetic variation for association, an intractable problem, we can leverage naturally occurring haplotyes to guide the analysis. Hidden Markov models have traditionally been used to model haplotypes and we can leverage HMMs in this setting as well in order to calculate the probability that a region of the genome is associated with a phenotype of interest.

One interesting investigation to run with this is to see if GWAS hits line up with regions identified by HapWAS.  If there are multiple casual variants and complex architectures, than a SNP that tags them may not be located exactly where they are, so the question is, is a HapWAS region centered more accurately on the causal variation?