78 2 OCTOBER 2020•VOL 370 ISSUE 6512 sciencemag.org SCIENCE
Fig. 3. Mutation burden and signatures in normal urothelium.(A) Scatterplot
of donor age versus the median number of substitutions in high-coverage
exomes (≥40× for≥80% of the exome). The fitted line,R^2 (coefficient of
determination) value, andPvalue were obtained by linear regression.
(B) Comparison of mutation burden between normal bladder urothelium and
bladder cancers. To account for subclonality, both a mean lower-bound estimate
per cell ( 22 ) and the mean number of mutations per microbiopsy are shown
for whole genomes from the 15 transplant organ donors. Bladder cancer
data reflect total mutations per genome from PCAWG ( 48 ). (C) Raw mutational
spectra for all urothelial genomes combined for three donors. (D) Number
(top) and proportion (bottom) of mutations assigned to the four most abundant
signatures extracted using a Bayesian hierarchical Dirichlet process ( 22 ) for
urothelial genomes from transplant organ donors. The weak attribution of
signature C (SigC) to genomes from T08 may reflect overfitting to residual
ATT > AAT alignment errors. (EtoH) Bar plots depicting mutational
spectra, split by type and trinucleotide context, of extracted signatures,
as in ( 17 ). (I) Intermutational distance plots for urothelial clones free from
and affected by APOBEC activity, respectively, as in ( 17 ). (J) Histology image
depicting variability in mutational processes between nearby urothelial
microbiopsies. Mutational spectra are from independent clones. (K) Fraction
of exomes with evidence of APOBEC mutagenesis ( 22 ). Error bars depict
95% binomial confidence intervals. (L) Proportion of exomes from normal
urothelium with large-scale copy number alterations in autosomes ( 22 ). Gains
(red) and losses (blue) are shown above and below thexaxis respectively.
(M) Copy number plots for representative whole genomes of normal
urothelium (top) and CIS (bottom).RESEARCH | RESEARCH ARTICLES