Scoring canonical driver genes

Hi there, great ML approach, easy to implement. I wanted to ask you whether (and perhaps it might require me to modify the codes or generate new ones) but is it possible to get a ranking of the canonical drivers predicted for each sample? 


Thanks!