GetStabilityLm {GeneSelector}R Documentation

Stability measures for gene rankings

Description

Assesses stability of gene rankings by regressing the rankings of perturbed datasets on the ranking of the original datasets in a weighted manner. The idea is that if stability is high, the resulting regression models fit well.

Usage

GetStabilityLm(RR, decay = c("linear", "quadratic", "exponential"), 
                measure = c("wilks", "direct"), 
                scheme = c("rank", "pval"), alpha = 1, ...)

Arguments

RR An object of class RepeatRanking.
scheme Whether ranks (scheme="rank") or p-values (scheme="pval") should be used as basis of weighting.
decay argument controlling the weight decay for the weights used in the linear regression model. If decay=linear, then the weight of the s-th rank/p-value is 1/s, if decay=quadratic, then the weight is 1/s^2 and if decay=exponential, then the weight is exp(-s*alpha), where alpha is a tuning parameter specified via the argument alpha.
measure The stability measure to be computed. If measure="wilks", then a stability measure based on the Wilk's Lambda Test for multivariate linear regression models is used. If measure="direct", then the direct generalization of the univariate coefficient of determination to the multivariate case is used. The second approach can fail if there exists rankings of the perturbed datasets that are exactly equal.
alpha To be specified only if decay="exponential", s. also GetAlpha.
... Further arguments passed to lm.

Value

An object of class GetStabilityLm

Author(s)

Martin Slawski martin.slawski@campus.lmu.de
Anne-Laure Boulesteix http://www.slcmsr.net/boulesteix

References

Mardia, K.V., Kent, J.T., Bibby, J.M. (1979).
Multivariate Analysis Academic Press.

See Also

GetRepeatRanking, GetStabilityOverlap, RecoveryScore, GetAlpha

Examples

### Load toy gene expression data
data(toydata)
### class labels
yy <- toydata[1,]
### gene expression
xx <- toydata[-1,]
### get ranking 
ordT <- RankingTstat(xx, yy, type="unpaired")
### Generate Leave-One-Out
loo <- GenerateFoldMatrix(xx, yy, k=1)
### Repeat Ranking with t-statistic
loor_ordT <- GetRepeatRanking(ordT, loo)
### assess stability
stab_lm_ordT <- GetStabilityLm(loor_ordT, decay="linear")
### plot
plot(stab_lm_ordT )

[Package GeneSelector version 1.2.0 Index]