cvscore: Genome-wide association scan based on cross-validated...
In cvGWAS: Genome-wide association analysis based on predictive ability

Description Usage Arguments Details Value Author(s) References See Also Examples

The function performs genome-wide association scan based on predictive ability evaluated via cross validation.

1 2	cvscore(formula, data, nfolds = 5, seed = NA, verbose = TRUE, ...)

`formula`	Formula describing fixed effects to be used in analysis, e.g. y ~ a + b means that outcome (y) depends on two covariates, a and b. If no covariates used in analysis, please skip the right-hand side of the equation.
`data`	An (optional) object of `gwaa.data-class` or a data frame with outcome and covariates. To read data into the require object format or to know more, please refer to the `GenABEL-package`.
`nfolds`	Number of folds to be established during the cross validation.
`seed`	A positive integer to set seed for randomization so that the results can be exactly repeated. If `NA`, no seed is used.
`verbose`	A logic value to set whether more screen displayed output is shown during the procedure.
`...`	Other parameters to be passed.

The predictive ability is assessed by an $R^2$ value, which is the squared correlation coefficient between the predicted phenotypes and their true values for each variant. In each fold of the cross validation, such an $R^2$ value is calculated, then the final estimate of $R^2$ is given by the mean of all such values from the cross validation.

A list containing R2.est, R2.raw and R.raw which provide the estimated $R^2$ for each variant, the orginal $R^2$ values from cross validation, and the corresponding $R$ values.

Xia Shen

Shen X (2013). The missing heritability revealed in Arabidopsis thaliana. Submitted.

GenABEL-package, qtscore

## Not run: 
require(GenABEL)
data(srdta)
cvres <- cvscore(qt3 ~ sex, data = srdta)
r2 <- cvres$R2.est
r2[which(is.na(r2))] <- 0
plot(r2, xlab = 'SNP index', ylab = 'Predictive ability')

## End(Not run)