Flowchart of DGS algorithm. Driver and targets are identified in a three step process, as shown. Candidate drivers are firstly selected from genes/miRNAs that reside in copy-number altered (CNA) regions and filtered by various procedures, for example based on fold-change and consistency between expression level and copy number status. The rests of genes/miRNAs are candidate targets, which are grouped based on correlation with the candidate drivers. Correlation between all drivers and targets in each module is highlighted using sparse canonical correlation analysis (SCCA).