Method-specific characteristics for the identification of differentially expressed genes in breast cancer.

The proportion of genes predicted as under- or overexpressed in relation to the total number of measured genes and the corresponding means and standard deviations of the underlying measured log-ratios are summarized for each method based on the the breast cancer gene expression data set from [3]. The different methods were grouped into three categories according to their proportion of identified differentially expressed genes and the corresponding mean log-ratio columns. The rightmost column specifies the runtimes of the different methods required to analyze the data set. All methods except ChARM, MixMod, DSHMM and AR()-HMM() were run on the ADaCGH web-server [52] utilizing parallel computations (AMD Opteron 2.2 GHz CPUs with 6 GB RAM). The remaining methods were run on a standard laptop with Intel CPU T9500 2.6 GHz and 2 GB RAM.