Traditional evolutionary biologists firmly believing in random mutation write a misleading blog-post about our HbS mutation results

Professors Brian Charlesworth and Jerry Coyne recently chose to criticize our recent paper not in a peer-reviewed format but on Coyne’s blog, and without inviting us to respond. We are posting our response below.

                                                                                                                                                                                                                                                                              July, 5 2022

Dear Professors Charlesworth and Coyne,

Our new data not yet published is strongly consistent with the data already published, showing that the pattern is not going to go away. In addition, we have predicted the pattern and have already demonstrated it empirically with our published data. But let's start with some things from our paper that you have not mentioned.


As a short introduction, we have developed the first method able to measure mutation rates at the single mutation resolution (whereas all previous studies measured mutation rates as averages across many genomic positions). On the general level, we found that mutation rates at that resolution vary much more and in a different manner than expected by common wisdom. For example, mutation rates varied substantially between genes and between populations even for the same mutations on the same local genetic background, suggesting that mutation rates are mutation specific and determined in a complex fashion, as predicted by our theory and not by the random mutation view. Now let's get more specific.


First, we found that the overall point mutation rate in the beta-globin region studied is significantly higher in Africans than in Europeans even while taking into account any possible amount of clonal dependence between sperm cells. Since this difference applies to a tiny (6 bp) region of great importance for adaptation, it already does not conform to the notion of random mutation as you define it.


Second, we found a correspondence between mutation rates at the single-mutation resolution and allele frequencies in populations. For example, the HbS mutation, which has been observed in the past on multiple different genetic backgrounds in Africans, originated in our samples multiple independent times de novo in Africans, whereas the Hb-Leiden mutation, which has been observed in the past on multiple genetic backgrounds across continents, originated in our samples multiple independent times de novo in both Africans and Europeans.

Third, we found that the HbS (20A>T) mutation originates significantly more frequently in HBB in sub-Saharan Africans compared to the other three cases combined (HBB in Europeans and HBD in either sub-Saharan Africans or Europeans), even while taking into account any possible clonal dependence between sperm cells. This result shows that the HbS mutation originates more frequently where it is of adaptive significance, even while fully accounting for your clonal dependence concern, and therefore it is inconsistent with the notion of random mutation as you define it, whether this result is due to a gene effect or to a population effect or both.


Fourth, the HbS effect actually applies independently to the gene and independently to the population and of course to both together. 1) Because the overall point mutation rate in the HBB region of interest (ROI) in Africans is significantly higher than that in Europeans (with or without counting the HbS mutation itself) while taking into account any possible clonal dependence between sperm cells, to say that the de novo HbS mutation rate is not higher in Africans than in Europeans is to say that you expect this mutation in particular to violate the statistically significant general pattern, when in fact it is already pointing strongly in the same direction, in a sense more so than any other single mutation in the region. 2) The HbS mutation rate also strongly conforms with the general and significant pattern of correspondence between de novo mutation rates at the single mutation resolution and past observations of alleles in carriers. Also on this additional account, to say that the HbS mutation rate does not differ between the populations is to say that that this mutation specifically must violate yet another significant general pattern, when in fact it is already pointing strongly in the same direction. Overall from points 1 and 2 above, by saying that there is no difference in the HbS rate between Africans and Europeans, you are actually expecting the HbS mutation rate to violate two independent statistically significant trends, both of which it already strongly conforms with. Rationally, this makes no sense. The onus of proof has obviously turned, so that it is now on those who think, counter this strongly supportive data, that there is no difference, to prove their now strange claim. Beyond any doubt, the data has reversed the expectation.


All that being said, the assumption you rely on that the instances within a donor are dependent (clonal dependence) is not called for given the data. Once statistically significant differences exist even under the assumption of maximal clonal dependence (such as the difference in the overall point mutation rates in the tiny region studied between the populations), then they require a biological explanation. But unless you wish to assume that a cellular-level mechanism induces specific mutations in a population-specific manner in accord with the cellular generation during spermatogenesis, it follows that any amount of clonal dependence would have only added noise to the data that is already exhibiting the statistically significant patterns, and in that sense the assumption of clonal dependence is conservative to the finding, which makes it unlikely that a noticeable amount of clonal dependence exists in the data to begin with. And, in that case, it also becomes obvious that it is much easier to attribute the repetitions of mutations within donor to a high de novo rate. (Not that it matters; as explained, we conducted our analyses precisely in such manner that addresses even the assumption of maximum clonal dependence.)

Your blog as accessed on July 5 2022 does not communicate any of these and other points of interest to the readers, the upshot of all of which is that, if one looks at the data in a clear-headed and open-eyed manner, it contradicts the random mutation view as you define it in many ways, not just one. In science, it is necessary to consider the data in its entirety—to rationally put together the different pieces into a coherent whole.

Finally, by bringing up the argument that it is “extremely hard to think of a mechanism” that would be responsible for these patterns you seem to be implying that the data could have possibly shown what you say it does not except that, given the lack of a mechanism for such a pattern given that mutation is random, the pattern must not be real. However, that seems to be putting the cart before the horse: it is the theory that must fit the data, not vice versa. I agree that it is extremely hard to imagine a mechanism that could explain our data when you look at things through the random mutation and natural selection lens. But that is why we need new ideas on how evolution happens. Interested readers are welcome to find a new scientific proposal, called Interaction-based Evolution (IBE), in our previous papers and preprints (see, for example, our online manuscript: Evolutionary honing in and mutational replacement: How long-term directed mutational responses to specific environmental pressures are possible), and note that our proposed explanation is fundamentally different from both of the basic traditional alternatives of mutations-as-errors and Lamarckism and should not be pigeonholed into old categories. 

In the meantime, thank you for having clarified that an increased de novo rate of the HbS mutation in Africans goes counter the notion of random mutation as you define it—counter the fundamental meaning of “random mutation”—but that you do not think that our data shows any such increase. New data that we have is strongly consistent so far with the results that we have already published, and interested readers are welcome to follow our future papers and data as well as data that others who are sparked by our results will bring. We will be eager to see how it all plays out.

Adi Livnat