Supplementary Figure 1: Tumor purity, mutational burden, and response to immune checkpoint therapy.

a, Called mutations by estimated tumor purity (ABSOLUTE) in 297 tumors with adequate sequencing quality (i.e., not excluded owing to sample contamination, low sequencing coverage, or other quality issues). Fifteen samples with high purity and outlying high mutational burdens are not shown. Purity <10% was associated with decreased ability to call mutations in most samples. The two samples with purity <10% and >500 called mutations/exome had unusually high sequencing coverage. However, inference of clonal versus subclonal mutational architecture was not possible by ABSOLUTE given low confidence in calling somatic CNAs, and a high rate of false negative mutation calls may have occurred, so these samples were excluded from analysis. Box plots show the median, first and third quartiles, whiskers extend to 1.5 times the interquartile range, and outlying points are plotted individually. b, Bar plot showing the frequency of estimated tumor purities, colored by patient response status. Tumor purity did not strongly predict clinical benefit from immune checkpoint therapies in 249 tumors included in the final analysis. c, Scatterplot showing subclonal mutational burden by estimated tumor purity in 249 included samples. Even after excluding tumors with estimated purity <0.1, subclonal mutations were likely undercalled in tumors with purity <0.2.