Real data: Application to C-to-U conversion data and application to nutrigenomic study

Two well-known real data sets were analyzed. The first was a binary classification problem, where the predictors were of different types. In the second, the response was multivariate and the predictors were continuous and categorical, as in the first set.

The first data set was the Arabidopsis thaliana, Brassica napus, and Oryza sativa data from [14, 15], which can be downloaded from Additional file 2. It applies to C-to-U conversion data. RNA editing is the process whereby RNA is modified from the sequence of the corresponding DNA template [14, 15]. For example, cytidine-to-uridine (C-to-U) conversion is usual in plant mitochondria. Although the mechanisms of this conversion are not known, it seems that the neighboring nucleotides are important. Therefore, the data set is formed by 876 cases, each of them with the following recorded variables (one response and 43 predictors):

Edit, with two values (edited or not edited at the site of interest). This is the binary response variable.

The 40 nucleotides at positions –20 to 20 (named with those numbers), relative to the edited site, with 4 categories;

the codon position, cp, which is categorical with 4 categories;

the estimated folding energy, fe, which is continuous, and

the difference in estimated folding energy (dfe) between pre-edited and edited sequences, which is continuous.

The second data set derives from a nutrigenomic study [16] and is available in the R package randomForestSRC [20] and Additional file 2. The study examines the effects of 5 dietary treatments on 21 liver lipids and 120 hepatic gene expressions in wild-type and PPAR-alpha deficient mice. Therefore, the continuous responses are the lipid expressions (21 variables), while the predictors are the continuous gene expressions (120 variables), the diet (categorical with 5 categories), and the genotype (categorical with 2 categories). The number of observations is 40. According to [16], in vivo studies were conducted under European Union guidelines for the use and care of laboratory animals and were approved by their institutional ethics committee.