Human breast cancer is a heterogeneous disease in terms of molecular alterations, cellular composition, and clinical outcome. Breast tumours can be classified into several subtypes, according to levels of mRNA expression (Sørlie et al., 2001 [1]). Here we consider a subset of data generated by The Cancer Genome Atlas Network (Network et al., 2012 [2]). Data were normalized and drastically pre-filtered for illustrative purposes. The data were divided into a training set with a subset of 150 samples from the mRNA, miRNA and proteomics data, and a test set includes 70 samples, but only from the mRNA, miRNA and methylation data (proteomics missing).

The mixOmics TCGA dataset is accessed via breast.TCGA' and contains the following:

breast.TCGA$data.train$mirna (continuous matrix): 150 rows and 184 columns. The expression levels of 184 different sections of miRNA.
breast.TCGA$data.train$mrna (continuous matrix): 150 rows and 200 columns. The expression levels of 200 different sections of mRNA.
breast.TCGA$data.train$protein (continuous matrix): 150 rows and 142 columns. The abundance of 142 different proteins
breast.TCGA$data.train$subtype (categorical vector): length of 150. Indicates the breast cancer subtype of each subject. Includes Basal, Her2 and LumA.