riskScore: Function to compute genetic risk scores.
In PredictABEL: Assessment of Risk Prediction Models

Description Usage Arguments Details Value Note See Also Examples

The function computes unweighted or weighted genetic risk scores. The relative effects (or weights) of genetic variants can either come from beta coefficients of a risk model or from a vector of beta coefficients imported into R, e.g., when beta cofficients are obtained from meta-analysis.

1	riskScore(weights, data, cGenPreds, Type)

`weights`	The vector that includes the weights given to the genetic variants. See details for more informations.
`data`	Data frame or matrix that includes the outcome and predictors variables.
`cGenPreds`	Column numbers of the genetic variables on the basis of which the risk score is computed.
`Type`	Specification of the type of risk scores that will be computed. Type can be weighted (`Type="weighted"`) or unweighted (`Type="unweighted"`).

The function calculates unweighted or weighted genetic risk scores. The unweighted genetic risk score is a simple risk allele count assuming that all alleles have the same effect. For this calculation, it is required that the genetic variables are coded as the number of risk alleles. Beta coefficients are used to determine which allele is the risk allele. When the sign of the beta coefficient is negative, the allele coding is reversed. The weighted risk score is a sum of the number of risk alleles multiplied by their beta coefficients.

The beta coefficients can come from two different sources, either beta coefficients of a risk model or a vector of beta coefficients imported into R, e.g., when beta cofficients are obtained from meta-analysis. This vector of beta coefficients should be a named vector containing the same names as mentioned in genetic variants. A logistic regression model can be constructed using fitLogRegModel from this package.

The function returns a vector of risk scores.

When a vector of beta coefficients is imported, it should be checked whether the DNA strands and the coding of the risk alleles are the same as in the study data. The functions are available in the package GenABEL to accurately compute risk scores when the DNA strands are different or the risk alleles are coded differently in the study data and the data used in meta-analysis.

plotRiskDistribution, plotRiskscorePredrisk

# specify dataset with outcome and predictor variables
data(ExampleData) 
# specify column numbers of genetic predictors
cGenPred <- c(11:16)

# fit a logistic regression model
# all steps needed to construct a logistic regression model are written in a function
# called 'ExampleModels', which is described on page 4-5
riskmodel <- ExampleModels()$riskModel2

# compute unweighted risk scores 
riskScore <- riskScore(weights=riskmodel, data=ExampleData, 
cGenPreds=cGenPred, Type="unweighted")