Response-related filtering of a synthetic data set. (a) The structure of the data set. The columns represent samples and the rows represent variables. (b) The projection score as a function of log10(α), where α is the p-value threshold used as inclusion criterion. (c) The projection score as a function of the cardinality of the variable subset. (d) The projection of the samples onto the first principal component from the most informative variable subset, consisting of 40 variables.