Figure 2From: TOPS: a versatile software tool for statistical analysis and visualization of combinatorial gene-gene and gene-drug interaction screensCorrelation of p-values for linear and split linear model algorithms. Splitting the data set into smaller subsets and normalizing as well as analyzing each of them separately only introduces minor deviations in the resulting candidate hits (i.e. 39 of the top 50 hits in one method are among the respective top 50 of the other method).Back to article page