standardizeaddon {bapred}

Addon batch effect adjustment for standardization

Description

Performs addon batch effect adjustment for standardization: 1) takes the output of standardize applied to a training data set together with new batch data; 2) checks whether the training data was also adjusted using standardization and whether the same number of variables is present in training and new data; 3) performs standardization on the new batch data.

Usage

standardizeaddon(params, x, batch)

Arguments

params

object of class standardize.

x

matrix. The covariate matrix of the new data. Observations in rows, variables in columns.

batch

factor. Batch variable of the new data. Currently has to have levels: '1', '2', '3' and so on.

Value

The adjusted covariate matrix of the test data.

Note

It is not recommended to perform standardization in cross-study prediction settings, because for some classifiers the prediction performance can be strongly impaired by this method. Given a not too small test set, the following methods are recommended (Hornung et al., 2016a): combatba, meancenter, ratioa, ratiog.

Because standardization is performed "batch by batch" the "addon procedure" for standardization consists of plain standardization on the new test batches.

Author(s)

Roman Hornung

References

Hornung, R., Causeur, D., Bernau, C., Boulesteix, A.-L. (2017). Improving cross-study prediction through addon batch effect adjustment and addon normalization. Bioinformatics 33(3):397–404, <doi: 10.1093/bioinformatics/btw650>.

Hornung, R., Boulesteix, A.-L., Causeur, D. (2016). Combining location-and-scale batch effect adjustment with data cleaning by latent factor adjustment. BMC Bioinformatics 17:27, <doi: 10.1186/s12859-015-0870-z>.

Examples

data(autism)

trainind <- which(batch %in% c(1,2))

Xtrain <- X[trainind,]
ytrain <- y[trainind]
batchtrain <- factor(as.numeric(batch[trainind]), levels=c(1,2))


testind <- which(batch %in% c(3,4))

Xtest <- X[testind,]
ytest <- y[testind]

batchtest <- as.numeric(batch[testind])
batchtest[batchtest==3] <- 1
batchtest[batchtest==4] <- 2
batchtest <- factor(batchtest, levels=c(1,2))


params <- standardize(x=Xtrain, batch=batchtrain)

Xtestaddon <- standardizeaddon(params=params, x=Xtest, batch=batchtest)