Fig. 2

Changing MutSigCV to MutSigCVsyn to identify genes that are enriched for synonymous mutations. A MutSigCV accounts for mutation heterogeneity across patients, diseases, and genes. B Comparison between MutSigCV and MutSigCVsyn: (1) MutSigCVsyn uses only non-coding mutations instead of a background comprised of both non-coding and synonymous mutations adopted by a majority of driver mutation detection algorithms. (2) MutSigCVsyn utilizes whole genome sequencing input data instead of whole-exome sequencing. (3) Both MutSigCVsyn and MutSigCV only utilize mutations in transcriptionally expressed regions. (4) MutSigCVsyn utilizes a re-annotated covariate file that was adapted to the PCAWG Gencode v19 annotation. (5) MutSigCVsyn patients have high-quality coverage data over non-coding regions, compared to limited coverage in the original MutsigCV. (6) MutSigCVsyn utilizes a non-parametric empirical Bayesian method to calculate the local FDR value. C The outline of MutSigCVsyn. Boxes with solid lines show MutSigCVsyn exclusive input/steps (see Methods section for detailed description)