QTDT - Tour | Previous

Population Stratification and The Total Evidence for Association

QTDT can evaluate the evidence for population stratification using a quantitative phenotype and a single marker (for details, see the papers of Fulker et al and Abecasis et al). Use the -ap model of association to do this. For example, qtdt -d sibs.dat -p sibs.ped -i sibs.ibd -x-99.999 -ap -wega evaluates the evidence for stratification at each marker in the sibs data set. Try it...

Testing trait:                          Trait
=============================================
Testing marker:                         SNP_1
---------------------------------------------
 Allele   df(0)  LnLk(0)   df(S)  LnLk(S)   ChiSq       p
    1 :     194   674.61     193   674.58    0.06          ( 164/200 probands)
Testing marker:                         SNP_2
---------------------------------------------
 Allele   df(0)  LnLk(0)   df(S)  LnLk(S)   ChiSq       p
    1 :     194   678.35     193   678.30    0.10          ( 152/200 probands)
Testing marker:                         SNP_3
---------------------------------------------
 Allele   df(0)  LnLk(0)   df(S)  LnLk(S)   ChiSq       p
    1 :     194   682.76     193   682.61    0.30          ( 168/200 probands)

In this data set, QTDT finds no evidence for stratification. If you are confident there is no stratification, you might test the total evidence for association (ie, not a TDT). To do this, use the -at option. For example, run qtdt -d sibs.dat -p sibs.ped -i sibs.ibd -x-99.999 -at -wega to evaluate the total evidence for association in the sibs data set.

Testing trait:                          Trait
=============================================
Testing marker:                         SNP_1
---------------------------------------------
 Allele   df(0)  LnLk(0)   df(X)  LnLk(X)   ChiSq       p
    1 :     195   685.20     194   674.61   21.18  0.0000  (200 probands)
Testing marker:                         SNP_2
---------------------------------------------
 Allele   df(0)  LnLk(0)   df(X)  LnLk(X)   ChiSq       p
    1 :     195   685.22     194   678.35   13.75  0.0002  (200 probands)
Testing marker:                         SNP_3
---------------------------------------------
 Allele   df(0)  LnLk(0)   df(X)  LnLk(X)   ChiSq       p
    1 :     195   685.21     194   682.76    4.91  0.0268  (200 probands)

The total evidence for association is strongest for SNP_1. What next, functional studies?... I hope you found this tour useful, and happy gene hunting!