rDriver output of BRCA and its comparison with other methods.

<p>(A) the frequency (blue) and GERP score (red) distribution across the 528 mutations in TCGA BRCA data (the SIFT score is not shown due to missing value) (B) the association matrix between the mRNAs and the mutations, with brown dots representing non-zero association coefficients. (C) the total number of non-zero values in the association matrix column-wise, representing the likelihood of driver mutations. A few known driver mutation hotspots in <i>PIK3CA</i> and <i>TP53</i> are labelled (red text). (D) Receiver operator characteristic (ROC) curves comparing the sets of driver genes predicted by various programs against a set of 17 known cancer driver genes in the Cancer Gene Census. (E) ROC curves comparing the sets of mutations predicted by rDriver, frequency, GERP, Condel, SIFT and PolyPhen against a set of 42 known driver mutations.