Evaluating the evaluation of cancer driver genes

Significance Modern large-scale sequencing of human cancers seeks to comprehensively discover mutated genes that confer a selective advantage to cancer cells. Key to this effort has been development of computational algorithms to find genes that drive cancer based on their patterns of mutation in large patient cohorts. Because there is no generally accepted gold standard of driver genes, it has been difficult to quantitatively compare these methods. We present a machine-learning–based method for driver gene prediction and a protocol to evaluate and compare prediction methods. Our results suggest that most current methods do not adequately account for heterogeneity in the number of mutations expected by chance and consequently yield many false-positive calls, particularly in cancers with high mutation rate. Sequencing has identified millions of somatic mutations in human cancers, but distinguishing cancer driver genes remains a major challenge. Numerous methods have been developed to identify driver genes, but evaluation of the performance of these methods is hindered by the lack of a gold standard, that is, bona fide driver gene mutations. Here, we establish an evaluation framework that can be applied to driver gene prediction methods. We used this framework to compare the performance of eight such methods. One of these methods, described here, incorporated a machine-learning–based ratiometric approach. We show that the driver genes predicted by each of the eight methods vary widely. Moreover, the P values reported by several of the methods were inconsistent with the uniform values expected, thus calling into question the assumptions that were used to generate them. Finally, we evaluated the potential effects of unexplained variability in mutation rates on false-positive driver gene predictions. Our analysis points to the strengths and weaknesses of each of the currently available methods and offers guidance for improving them in the future.