http://www.cell.com/AJHG/abstract/S0002-9297(10)00204-1
If an association can be accounted for by multiple causal variants, each of which is likely to be rare, then some true causal variants could lie outside of the LD block containing the most significant tag SNPs, and multiple seemingly independent association signals may be present at the same locus.16 As a corollary, resequencing a narrow LD block at susceptibility loci in a small number of randomly selected control samples may not reveal the causal variants. Whole-genome imputation, even after the completion of the 1000 Genomes Project, may not find the causal variants either, unless rare variants (usually ethnicity-specific) are present in the haplotype data and can be imputed accurately.
There are a few known examples supporting the hypothesis that multiple rare variants may collectively be responsible for an association signal......In summary, none of these association signals on common tag SNPs discussed above are spurious signals; instead, they may represent scenarios whereby multiple causal variants work together to create genome-wide significant association signals, which are being accredited to one common tag SNP.
The purpose of the current study is not to speculate on the fraction of association signals that can be attributed to the presence of multiple causal alleles. Rather, our aim is to accept the possibility that some signals in GWAS emerge from rare causal variants, and to use this possibility to leverage the extensive GWAS data in the search for causal variants. On the basis of the observation of differential LD between tag SNPs in cases compared to controls, together with the observation of long-range haplotypes surrounding tag SNPs, we present an approach for selecting cases to maximize the chance of finding causal variants. We evaluated this approach on a simulated data set and compared it with conventional fine-mapping approaches to identify causal variants. We also tested the approach on a GWAS on hearing loss, in which we know the identity of the causal variants, and we have sequenced all available cases for the presence of causal variants. We believe that our theoretic framework and our case-selection approach will have significant implications for the design of follow-up studies after a successful GWAS in order to facilitate success in finding the causal genes and their causal variants.
http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0002504
The ‘Common Disease-Common Variant’ Hypothesis and Familial Risks
http://www.nature.com/nrg/journal/v5/n2/pdf/nrg1270.pdf
The purpose of the current study is not to speculate on the fraction of association signals that can be attributed to the presence of multiple causal alleles. Rather, our aim is to accept the possibility that some signals in GWAS emerge from rare causal variants, and to use this possibility to leverage the extensive GWAS data in the search for causal variants. On the basis of the observation of differential LD between tag SNPs in cases compared to controls, together with the observation of long-range haplotypes surrounding tag SNPs, we present an approach for selecting cases to maximize the chance of finding causal variants. We evaluated this approach on a simulated data set and compared it with conventional fine-mapping approaches to identify causal variants. We also tested the approach on a GWAS on hearing loss, in which we know the identity of the causal variants, and we have sequenced all available cases for the presence of causal variants. We believe that our theoretic framework and our case-selection approach will have significant implications for the design of follow-up studies after a successful GWAS in order to facilitate success in finding the causal genes and their causal variants.
http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0002504
The ‘Common Disease-Common Variant’ Hypothesis and Familial Risks
http://www.nature.com/nrg/journal/v5/n2/pdf/nrg1270.pdf
No comments:
Post a Comment