Reader Comments

Post a new comment on this article


Posted by bpadhuka on 18 Feb 2012 at 12:16 GMT

In the second model, a variant of the previous model, we assume that , where G1 is normally distributed with standard deviation σG1, and is Binomially distributed. In this case, Xi corresponds to SNPs with large effects and G1 represents many other small genetic effects; if there are enough small genetic effects, we expect that the asymptotic behavior of their sum would be according to a normal distribution. By setting the parameters λ, σG1 and p appropriately, we can control the relative risks of the large effect SNPs. We tune these parameters such that the relative risks are close to values observed in Table 1 (see below). As for the previous model, we can show that when G is known (but E is unknown) and the relative risks of the large effect SNPs and risk-allele frequencies are fixed, the area under the ROC curve for the second model only depends on the heritability and the average lifetime risk of the disease (see below).

This theoretical disease model can be made more general by allowing the genetic variable of a particular SNP to take any 3 arbitrary values corresponding to the 3 possible genotypes (instead of 0, LAMBDAi, and 2LAMBDAi as in the current variant for the NN, RN and RR genotypes respectively say X0(i), X1(i) and X2(i)).

No competing interests declared.


bpadhuka replied to bpadhuka on 26 May 2012 at 01:01 GMT

In addition, we can relax the Hardy-Weinberg equilibrium assumption for the genotype frequencies. See the following
link for more details about this model:

Competing interests declared: I am the first author of this paper.