Use of pre-transformation to cope with outlying values in important candidate genes

Outlying values in predictors often strongly affect the results of statistical analyses in high-dimensional settings. Although they frequently occur with most high-throughput techniques, the problem is often ignored in the literature. We suggest to use a very simple transformation, proposed before in a different context by Royston and Sauerbrei, as an intermediary step between array normalization and high-level statistical analysis. This straightforward univariate transformation identifies extreme values and reduces the influence of outlying values considerably in all further steps of statistical analysis without eliminating the incriminated observation or feature. The use of the transformation and its effects are demonstrated for diverse univariate and multivariate statistical analyses using nine publicly available microarray data sets. R-codes for reproducing the whole analysis are available from the companion website: http://www.ibe.med.uni-muenchen.de/organisation/mitarbeiter/020−professuren/boulesteix/outliers/, such that the study is completely reproducible. ∗Corresponding author. Email: [email protected]. 1