Experimental results

In this section we first report the parameters used to simulate the four real gene expression data sets Colon [3], Leukemia [7], DLBCL-FL and DLBCL-outcome [12], by using the procedure derived from the mathematical model, and the characteristics of "raw" gene expression data generated according to simple Gaussian distributions. In the main paper we show that Colon-cancer-like, Leukemia-like and DLBCL-like data are likely to be generated from the same distribution of real gene expression data, while "raw" values significantly differ from real ones, according to the Diggle statistical test.

In the second part of this section we present the comparison of the results obtained with real and simulated data using five different gene selection methods. The results are reported for each of the four considered real data sets: Colon, Leukemia, DLBCL-FL and DLBCL-outcome.

In the third part of this section we show the results of the comparison of two well-known gene selection methods proposed in the literature, to exemplify the application of the mathematical model to the assessment of the effectiveness of gene selection methods.