As we age, many tissues become colonized by microscopic clones carrying somatic driver mutations1–7. Some of these clones represent a first step towards cancer whereas others may contribute to ageing and other diseases. However, our understanding of this phenomenon remains limited due to the challenge of detecting mutations in small clones. Here we introduce a new version of nanorate sequencing (NanoSeq)8, a duplex sequencing method with an error rate lower than five errors per billion base pairs, which is compatible with whole-exome and targeted capture. Deep sequencing of polyclonal samples with single-molecule sensitivity simultaneously profiles large numbers of clones, providing accurate mutation rates, signatures and driver frequencies in any tissue. Applying targeted NanoSeq to 1,042 non-invasive samples of oral epithelium and 371 blood samples from a twin cohort, we report an extremely rich selection landscape, with 46 genes under positive selection in oral epithelium, more than 62,000 driver mutations and evidence of negative selection in essential genes. High-resolution maps of selection across coding and non-coding sites are obtained for many genes: a form of in vivo saturation mutagenesis. Multivariate regression models enable mutational epidemiology studies on how exposures and cancer risk factors, such as age, tobacco or alcohol, alter the acquisition or selection of somatic mutations. Accurate single-molecule sequencing provides a powerful tool to study early carcinogenesis, cancer prevention and the role of somatic mutations in ageing and disease.
