[BioC] GWASTools: assocTestRegression, forcing the analysis to use reference allele

Stephanie M. Gogarten sdmorris at u.washington.edu
Tue Aug 12 20:43:34 CEST 2014

Hi Danica,

There is not a way to change which allele is being tested in the code, 
but it is fairly straightforward to adjust the output to compare with a 
different effect allele.  The output contains a "minor.allele" column 
which will tell you whether allele A or B was the minor (tested) allele 
in your sample.  If the minor allele was B but the reference allele was 
A, then you would reverse the sign of "Stat" (beta) and take the inverse 
of the odds ratio to get the effect in terms of allele A.

Defining which allele to test is a somewhat arbitrary choice, and since 
the definition of A and B alleles for GWAS data is usually determined by 
the array manufacturer, we chose to use the minor allele for greater 
consistency.  (This is the same convention followed by PLINK, a popular 
tool for association testing.)  Testing the reference allele would 
require additional information not contained in the dataset itself 
(which organism?  which version of the reference sequence?).


On 8/12/14 2:37 AM, Danica [guest] wrote:
> In the package GWASTools, the function assocTestRegression performs an association test based on the minor allele. However this makes comparison across different datasets complicated because the minor allele in one population may be different from another.
> Is there a way to perform this test using the reference allele instead?
> Thanks,
> Danica
>   -- output of sessionInfo():
> --
> Sent via the guest posting facility at bioconductor.org.

More information about the Bioconductor mailing list