Summary
Mutation accumulation varies across a genome by chromosomal location, nucleotide identity, surrounding sequence, and chromatin context1–⇓⇓⇓5. Nevertheless, while mutagens, replication machinery, and repair processes exhibit identifiable mutation signatures, at the tissue or organismal scale the aggregate somatic manifestation of these processes has been difficult to measure, and often appears to be semi-random. This randomness is often believed to contribute to the stochasticity of diseases like cancer6 and physiological decline during aging7. The challenge in observing any tissue-wide somatic mutation patterns is that prior to clonal expansion, most mutations are rare in healthy tissue8–⇓⇓11. Here we describe a new method called FERMI (Fast Extremely Rare Mutation Identification), which comprehensively captures and quantifies rare mutations at single DNA molecule resolution that exist at frequencies as rare as 10′4 in human peripheral blood. Using this method, we observed an unanticipated degree of ubiquity and similarity between the somatic mutation loads of different individuals, where most assayed substitutions are found to occur at conserved frequencies across nearly all individuals spanning a nine decade age range. These observed mutational patterns existed both within within non-conserved, non-coding and non-repetitive regions of the genome and within the coding regions of oncogenes implicated in hematopoietic malignancies. Furthermore, we find that nucleotides preferentially mutate to particular bases in a manner that is specified by nucleotide identity, position, and sequence context. Finally, we identify individuals who deviate from typical mutational patterns in a reproducible manner that resembles a mild mismatch repair deficiency, suggesting that variation in somatic mutation rates may be relatively common. This study provides one of the first estimates of mutation burden in terminally differentiated somatic cells and demonstrates that somatic mutations in such cells are significantly more frequent and deterministic than previously believed, and are governed by mechanisms that when perturbed, result in predictable outcomes.