SNPassoc (version 1.4-9)

interactionPval: Two-dimensional SNP analysis for association studies

Description

Perform a two-dimensional SNP analysis (interaction) for association studies with possible allowance for covariate

Usage

interactionPval(formula, data, quantitative = 
          is.quantitative(formula, data), model = "codominant")

Arguments

formula

a formula object. It might have either a continuous variable (quantitative traits) or a factor variable (case-control study) as the response on the left of the ~ operator and the terms corresponding to the

data

a required object of class 'setupSNP'.

quantitative

logical value indicating whether the phenotype (those which is in the left of the operator ~ in 'formula' argument) is quantitative. The function 'is.quantitative' returns FALSE when the phenotype

model

a character string specifying the type of genetic model (mode of inheritance). This indicates how the genotypes should be collapsed. Possible value are "codominant", "dominant", "recessive", "overdominant" or "log-additive". The d

Value

• The 'interactionPval' function returns a matrix of class 'SNPinteraction' containing the p values corresponding to the different likelihood ratio tests above describe. Methods defined for 'SNPinteraction' objects are provided for print and plot. The plot method uses 'image' to plot a grid of p values. The upper triangle contains the interaction (epistasis) p values from LRT. The content in the lower triangle is the p values from the LRT comparing the additive model with the best single model. The diagonal contains the main effects pvalues from LRT. The 'plot.SNPinteraction' function also allows the user to plot the SNPs sorted by genomic position and with the information about chromosomes as in the 'plotMissing' function.

Details

The 'interactionPval' function calculates, for each pair of SNPs (i,j), the likelihood underling the null model L0, the likelihood under each of the single-SNP, L(i) and L(j), the likelihood under an additive SNP model La(i,j), and the likelihood under a full SNP model (including SNP-SNP interaction), Lf(i,j). The upper triangle in matrix from this function contains the p values for the interaction (epistasis) log-likelihood ratio test, LRT, LRTij = -2 (log Lf(i,j) - log La(i,j)) The diagonal contains the p values from LRT for the crude effect of each SNP, LRTii = -2 (log L(i) - log L0) The lower triangle contains the p values from LRT comparing the two-SNP additive likelihood to the best of the single-SNP models, LRTji = -2 (log La(i,j) - log max(L(i),L(j))) In all cases the models including the SNPs are adjusted by the covariates indicated in the 'formula' argument. This method is used either for quantitative traits and dicotomous variables (case-control studies).

References

JR Gonzalez, L Armengol, X Sole, E Guino, JM Mercader, X Estivill, V Moreno. SNPassoc: an R package to perform whole genome association studies. Bioinformatics, 2007;23(5):654-5.

See Also

setupSNP

Examples

data(SNPs)
datSNP<-setupSNP(SNPs,6:40,sep="")

ansCod<-interactionPval(log(protein)~sex,datSNP)
print(ansCod)
plot(ansCod)