The article by Shendure and Akay (2015) is the only one that addresses human mutation rates in any meaningful way. They begin their review with ...
Despite the exquisite molecular mechanisms that have evolved to replicate and repair DNA with high fidelity, mutations happen. Each human is estimated to carry on average ~60 de novo point mutations (with considerable variability among individuals) that arose in the germline of their parents (1–4). Consequently, across all seven billion humans, about 1011 germline mutations—well in excess of the number of nucleotides in the human genome—occurred in just the last generation (5). Furthermore, the number of somatic mutations that arise during development and throughout the lifetime of each individual human is potentially staggering, with proliferative tissues such as the intestinal epithelium expected to harbor a mutation at nearly every genomic site in at least one cell by the time an individual reaches the age of 60 (6).The main point in this paragraph is important. No matter what the exact value of the human mutation rate, every single possible point mutation will happen in just a few generations somewhere among the seven billion or so people on Earth. And each individual who lives to the ripe old age of 60 (i.e. youngsters) will have experienced a huge number of somatic mutations. Hundreds, possibly thousands, of these can cause cancer.
But let's think about the mutation rate. The value of ~60 new point mutations per generation in Shendure and Akay (2015) is quite a bit lower than the number I prefer (>100). My estimate is based on my understanding of the scientific literature. I've discussed this before in Human mutation rates - what's the right number?. The discussion in the comments section of my post is very enlightening; it reveals that the issue is complicated and all estimates, by any method, are open to criticism.
Shendure and Akay gave us four references to support their claim that the rate is ~60 mutations per generation.
- The first reference is paper I had not read. The authors (Campbell et al, 2012) report the genome sequences of five trios in the Hutterite population and conclude that the mutation rate is 76 mutations per generation. This is similar to the value obtained by other direct sequencing methods and subject to the same qualifications.1
- I also hadn't read the second paper. Francioli et al. (2015) sequenced the parents and offspring of 258 Dutch families and found a total of 11,020 de novo mutations. This corresponds to 43 mutations per generation. The paper frequently mentions "mutation rate" but they are referring to their own data and the number of mutations per generation that they detect. On first glance, I don't see how to correct for the amount of sequence they compared—was it about half of the reference genome sequence, in which case the real mutation rate would be 83 mutations per genome?
- The third paper is Roach et al. (2010). This is also a sequencing paper but one that I've discussed previously. The authors sequenced parents and two children and arrived at an error rate of 70 new mutations per generation.
- The fourth paper is Kong et al. (2012). It's the one that looked at genome sequences of Icelanders with fathers of various ages and concluded that the number of mutations went up as the age of the father increased. They estimate that the average mutation rate is 76 mutations per generation.
All of the rates are based on genomic sequencing. I'm not sure how you get from three estimates of 76, 76, and 70 to the Shendure and Akay estimate of ~60 point mutations per generation. Perhaps they misinterpreted the data of Francioli et al. (2015)?
Shendure and Akay (2015) discuss a number of different ways of calculating mutation rates. The first is based on the rates of mutation causing genetic diseases. These estimates date back to the 1920s and they are the reason why genetic load arguments were developed around 1950.
Typical values for genetic diseases were about 2 × 10-5 per gene per generation. It's almost impossible to relate this to an overall average mutation rate. There are more modern estimates that may be helpful according to Shendure and Akay ...
A number of distinct approaches have been used to estimate the germline mutation rate of base substitutions, which we focus on here unless otherwise noted. Historically (18), and even more recently (6), estimates of mutation rates have been derived from the incidence of highly penetrant Mendelian diseases. The largest such study aggregated data across ~60 loci, estimating an average germline mutation rate of 1.28 × 10−8 per base pair (bp) per generation (6) [82 mutations per generation].Reference 18 is a 2004 reprint of a 1935 paper by J.B.S. Haldane. A much more appropriate reference would be Haldane's 1949 review (Haldane, 1949) where he discusses all of the problems associated with these studies. Haldane concludes that typical mutation rates are about 2 × 10-5 per gene per generation.
Reference 6 is Michael Lynch's 2010 PNAS paper that looks at a database of deleterious mutations causing genetic diseases in humans (Lynch, 2010). Lynch is mostly interested in the somatic cell mutations rate because that's what causes the most problems. He concludes that the somatic cell mutation rate is about 8 × 10-10 per site per cell division and this is quite a bit higher that the germline mutation rate (about 1 × 10-10).
The somatic cell mutation rate gives rise to thousands of mutations per cell by the time an individual reaches maturity. Lynch notes that a high somatic cell mutation rate imposes an enormous mutational load on humans.
With respect to germline mutations, Lynch calculates that the average newborn has about 76 de novo point mutations although he points out that there will also be a number of new small deletions and insertions.
Shendure and Akay know that calculating germline mutation rates from genetic diseases is very difficult. They discuss how comparisons of DNA sequences from different species can be helpful.
Phylogenetic methods have also been used to estimate mutation rates at putatively neutral loci on the basis of the amount of sequence divergence between humans and nonhuman primates, yielding a higher genome-wide average germline mutation rate of 2.2 × 10−8 per bp per generation (19). Phylogenetic methods also make assumptions such as the time to most recent common ancestor between humans and nonhuman primates, generation time, and that the loci studied do not have fitness consequences. In addition, phylogenetic estimates may be influenced by evolutionary processes other than mutation and selection, such as biased gene conversion, which influences substitution rates in mammals (20).Reference 19 is the chimpanzee genome sequence paper from 2005. It's just one of many papers that use the phylogeny method to estimate a mutation rate [see Estimating the Human Mutation Rate: Phylogenetic Method]. Reference 20 is a 2008 paper on biased gene conversion. It's not likely that biased gene conversion will have a serious effect on calculating a germline mutation rate by the phylogenetic method. There are other, more important, sources of error.
A mutation rate of 2.2 × 10−8 per bp per generation correspond to 140 new mutations in every newborn baby (actually, in every zygote). This is about twice the rate determined by the direct sequencing method [see Estimating the Human Mutation Rate: Direct Method].
There's another way of estimating the mutations rate and that's the biochemical method [Estimating the Human Mutation Rate: Biochemical Method]. It's based on our understanding of the error rates of DNA replication and repair and it yields a value of 130 mutations per generation. This is closer to the phyogenetic method than to the direct sequencing method. Shendure and Akay don't discuss this method.
Shendure and Akey obviously prefer the estimates derived from direct sequencing of parents and children although they've picked a value (60 mutations per generation) that's on the low side of all those estimates. Nevertheless, they present a fair summary of the problems with all these estimates.
New sequencing technologies have enabled more direct estimates of mutation rates by identifying de novo mutations in pedigrees (i.e., those observed in a child but not their parents). Whole-genome sequencing studies (1–3, 9–11) of pedigrees estimate the germline mutation rate to be ~1.0 × 10It is not correct to say that the direct sequence estimates are "broadly consistent" with the analysis of Neanderthal and Denisovan genomes. In fact, a low estimate (e.g. 60 mutations per generation) would place the common ancestor of humans and chimps at 10 million years ago and the divergence of Neanderthals and Denisovans from the modern human lineage would be pushed back to unreasonable times [see Human mutation rates - what's the right number?].
−8 per bp per generation [extensively reviewed in (8)], which is less than half that of phylogenetic methods but in better agreement with disease-based estimates. An important caveat to pedigree-based sequencing is that heavy data filtering is necessary and analysis choices may influence both false-positive and false-negative rates (8). Nonetheless, complementary approaches for estimating mutation rates on the basis of the number of “missing mutations” that would be expected to have occurred in the time between when an archaic hominin individual (such as Neanderthal) died and the present (21) and the accumulation of heterozygous variants within autozygous regions of founder populations (1) are broadly consistent with pedigree-based approaches (~1.1 and 1.2 × 10−8 per bp per generation, respectively).
A twofold difference has implications so let's hope this debate is resolved soon. If we consider the upper estimates by direct sequencing then we approach 100 mutations per generation and that's not too much different than the values calculated by the phylogenetic method and the biochemical method.
I don't think that estimating human mutation rates from the frequency of human genetic diseases is anywhere near as accurate as the other methods.
1. There is a huge background of unique differences in every genome due to sequencing errors and artifacts. The only way to identify new mutations in offspring is to use sophisticated algorithms to eliminate false positives and false negatives. The number of true mutations is one or two orders of magnitude less that the noise due to errors.
Campbell, C.D., Chong, J.X., Malig, M., Ko, A., Dumont, B.L., Han, L., Vives, L., O'Roak, B.J., Sudmant, P.H., and Shendure, J. (2012) Estimating the human mutation rate using autozygosity in a founder population. Nature Genetics, 44:1277-1281. [doi: 10.1038/ng.2418]
Francioli, L.C., Polak, P.P., Koren, A., Menelaou, A., Chun, S., Renkens, I., van Duijn, C.M., Swertz, M., Wijmenga, C., van Ommen, G., Slagboom, D.I., Ye, K., Guryev, V., Arndt, P.F., Kloosterman, W.P., de Bakker, P.I.W., and Sunyaev, S.R. (2015) Genome-wide patterns and properties of de novo mutations in humans. Nature Genetics. 47:822-826. [doi: 10.1038/ng.3292]
Haldane, J. (1949) The rate of mutation of human genes. Hereditas, 35(S1):267-273. [doi: 10.1111/j.1601-5223.1949.tb03339.x]
Kong, A., Frigge, M.L., Masson, G., Besenbacher, S., Sulem, P., Magnusson, G., Gudjonsson, S.A., Sigurdsson, A., Jonasdottir, A., and Jonasdottir, A. (2012) Rate of de novo mutations and the importance of father/'s age to disease risk. Nature, 488:471-475. [doi: 10.1038/nature11396]
Lynch, M. (2010) Rate, molecular spectrum, and consequences of human mutation. Proceedings of the National Academy of Sciences, 107:961-968. [doi: 10.1073/pnas.0912629107
Roach, J.C., Glusman, G., Smit, A.F.A., Huff, C.D., Hubley, R., Shannon, P.T., Rowen, L., Pant, K.P., Goodman, N., and Bamshad, M. (2010) Analysis of genetic inheritance in a family quartet by whole-genome sequencing. Science, 328:636-639. [doi: 10.1126/science.1186802]
Shendure, J., and Akey, J.M. (2015) The origins, determinants, and consequences of human mutations. Science, 349:1478-1483. [doi: 10.1126/science.aaa9119]
The Chimpanzee Sequencing and Analysis Consortium (2005) Initial sequence of the chimpanzee genome and comparison with the human genome. Nature, 437:69-87. [doi: 10.1038/nature04072]