Summary of the overall comparative performance of FusionFinder, FusionMap and Tophat-Fusion on a simulated dataset.

A total of 55 fusion genes were simulated. Sensitivity and PPV measures were compiled from predicted fusion genes evidenced by 1 read or more (i.e. all data). True positive fusions/reads relate to the accurate prediction of simulated fusions whereas false positive fusions/reads relate to all other predictions including synonymous fusions. For FusionFinder, all false positive genes and consequently all false positive reads were from synonymous fusions (see Table S2b). FusionFinder detects more simulated fusions and significantly fewer false positives than FusionMap with consistently greater sensitivity and PPV. FusionFinder showed a higher sensitivity and comparable PPV to Tophat-Fusion.