Figure 5

Scatter plots of tumor vs normal variant allele fractions, using deep-seq pairs (A), exome-seq pairs (B), and tumor RNA-seq and normal exome-seq (C) from 39 LUSC patients. Each point is a variant site detected in the tumor exome-seq data using the GATK UnifiedGenotyper. Variants detected within 76 genes (targeted regions of the deep-sequencing data) are aggregated over the 39 patients (in total, 6,692). Using the deep-sequencing data, we determined the validation status as ‘somatic’ if the tumor vaf is > 10% and the normal vaf is < 2% (334 sites; red). Among the ‘non-somatic’ ones, we further classified the variants as ‘wildtype’ if the tumor vaf is < 2% and the normal vaf is < 2% (319 sites; green).