mtDNA molecular clock not real? Wallace's reply and my rebuttal

Dear Professor Huang:

Thank you for your discussion about mtDNA sequence divergence times. It is true, that because of the effects of selection, the entire mtDNA tree cannot be considered to function as a linear molecular clock.  This was a concern in 1983 as it is now. In 1983, there was archeological data that was felt to be compelling showing an Asian origin of Homo sapiens.   However, this was inconsistent with our mtDNA sequence divergence calculations.  As a result, we felt compelled to entertained two alternative explanations of the mtDNA tree, one based exclusively on mtDNA sequence divergence  and indicating an African origin and the other taking into account the then archeological data which required that the mtDNA sequence evolution rate would have had to be different between Africa and Asia.  

Between 1983 and now, we have obtained thousands of complete mtDNA sequences and identified many of the likely mtDNA variants that have been acted on by positive selection.  These unique variants create nodes in the mtDNA tree at the base of regional mtDNA lineages (haplogroups).  These nodes must be taken into account when calculating genetic distances.  Between the nodes are periods of the accumulation of seemingly "neutral" mutations which can be considered as acting in a clock-like fashion.

Having determined this, we concluded we could use the accumulation of mtDNA variation between nodes to calculate relative ages of regional populations.  This revealed that the amount of mtDNA variation that accumulated in African macrohaplogroup L was greater than that which has accrued within Eurasia macrohaplogroup N or Asian macrohaplogroup M.  Hence, we now feel that we can adjust for the potential effects of selection in creating a non-linear clock.  The conclusion is that the sequence divergence in macrohaplogroup L is much greater than macrohaplogroups M or N validating an out-of-Africa conclusion.

I hope this helps.

All the best,

Doug

Douglas C. Wallace, Ph.D.

Michael and Charles Barnett Endowed Chair in Pediatric Mitochondrial Medicine and Metabolic Disease

Director, Center for Mitochondrial and Epigenomic Medicine (CMEM)

Children's Hospital of Philadelphia

Professor, Department of Pathology and Laboratory Medicine

University of Pennsylvania

Dear Doug,

I am happy and surprised to receive your response almost immediately. I have learned a lot more from your response than reading many papers. I hope that I will be able to receive your insights on many important issues in our field.

I have been following my own interests and our own data in recent years. They have often led us to somewhat unconventional assumptions and conclusions. I would love to hear your comments on some of them so that we can carry out additional research.

After reading your response I have additional questions that I am hoping to hear your thoughts. The most striking thing to me upon reading your letter was that you have seemingly unknowingly admitted a fatal flaw in the field of molecular phylogeny that would qualify this field as it is presently practiced as, I am afraid to say, pseudoscience. For the field to be credible, researchers must be able to generate meaningful results based on logical deductions from well-defined assumptions with few references to non-molecular data except in the case of a few fossils serving as calibration for deriving substitution rates. Thus, whether modern humans originated in Asia or not should be a deduction of the molecular methods alone, regardless of any archaeological/fossil findings in Asia. Only in this double blind fashion, the molecular results could use the archaeological/fossil findings as confirmatory evidence and vice versa. If however, one fudges the parameters/assumptions of the molecular methods to meet a prior known non-molecular type finding, the molecular results would be meaningless. I am afraid that nothing could be more pseudoscientific than that.

On the other hand, it does seem like a routine practice in the field to constantly modify molecular parameters in order to fit a molecular result to an unexpected fossil find that challenged a previous molecular dating, while in the meantime never admitting any fundamental flaws in their methods/assumptions/theory that may be the reason for the constant rewriting in the first place. For example, the human-ape split was dated to be ~5 million years ago by Sarich and Wilson in 1967 using the molecular clock approach. However, the consensus now is 7 million years ago, which was forced upon us by recent fossil finds such as Sahelanthropus tchadensis. Now if future finds would put the human-ape split to even older times (there is this 9.7 million year old homo like teeth in Europe reported two weeks ago), would anyone insist on the much younger date derived by the clock method? It is almost certain that no one would. This just looks really bad for the molecular clock method as it is presently practiced: it has no credibility.

However, this is, in my opinion, definitely not inherently the best the molecular approach can do. We have in recent years developed a new way of doing this, the slow clock method, that we have found to be able to do the same job much better. Fossil finds that constantly push the homo-ape split into deeper times have essentially validated our dating of the event at ~18 million years ago or made it looking increasingly realistic (see Huang, 2012). And we definitely are not prepared to change our dating depending on what future fossil finds may say. We would not have published our dating if we were uncertain about it. Why we could be so certain is because our dating results were based on assumptions that are certain and solid, close to axioms if you will.  

That you can easily overlook the molecular clock assumption and its deduction the African Eve when faced with non-molecular findings pointing to Asia origin as you have admitted in your letter and in your 1983 paper just means that you have no real data to support the molecular clock and the African Eve. From such uncertain assumption like the molecular clock, any deduction from it, such as the African Eve, would also be equally uncertain. That such a deduction has been sold to the public and experts outside the field as fact is just unfortunate and sad.

You wrote that “Between the nodes are periods of the accumulation of seemingly "neutral" mutations which can be considered as acting in a clock-like fashion.” I am afraid I cannot agree with you on this because you did not provide any data and reasoning to back it up. Was it another deduction from uncertain assumptions or itself just an uncertain assumption? Even if there is a molecular clock, one still needs to exclude the possibilities of maximum saturation in genetic diversities and higher levels of tolerance of mutations in Africans in order to deduce the Out of Africa model. Those possibilities have yet to be addressed by the proponents of the Out of Africa model even though it is not very hard to do at all. And we did that and verified that genetic diversities in human populations today are at maximum saturation levels and Africans have greater levels of tolerance of mutations (I have been accused of certain bad things by some Internet comments for demonstrating this but facts do not lie and let’s not cheat ourselves and our students). The bottleneck assumption for non-Africans can be easily tested to be false as the true neutral variants as found by us showed no bottlenecks and very low level of sharing between Africans and non-Africans. Therefore, the Out of Africa model has no proper justifications even if I grant you the molecular clock. More on this see our preprint and references therein, Yuan et al 2017.

It is much easier to see what inherently went wrong with the molecular clock ‘mirage’ if one can provide an alternative perspective that does a much better job. If it is a mirage, then what is the real thing instead? We have been working in the past decade to try to offer exactly that. Briefly, we have shown that the original protein alignment results of Zuckerkandl/Pauling and Margoliash should never have been interpreted by the same mutation rate hypothesis, i.e., the molecular clock and in turn the Kimura and King/Jukes neutral theory that was inspired by the clock. See Kumar 2015 ‘four decades of molecular clock’ paper for a history on this. To use the equation r=d/2t to derive mutation rate requires one to know beforehand whether the d (distance) in question is at maximum saturation or not (maximum distance does not vary with time and must not be used for the calculation). But unfortunately the maximum distance question was never even raised and all 3 of these pioneers assumed that they were dealing with linear or near-linear distances that would always increase with time. We have however found that those distances they saw are all maximum distances. One can easily show this by counting the number of repeatedly mutated positions (more means saturation), Huang, 2010, 2016.

By the way, the present mtDNA tree requires knowing derived and ancestral alleles, which requires the assumption of no repeatedly mutated positions or infinite sites. But the fact is that nearly all observed variant positions show repeated mutations, clearly violating the infinite sites assumption. For example, the 4248 site mutation is supposed to be one of the mutations that define haplotype A but occurs also often in haplotype E. So the tree’s premises (infinite sites and no recurrent mutations) are grossly violated by the tree itself, which is a very awkward and stretching story to tell to the world, especially to people who are just being minimally scientifically sound and competent.

We have used our novel perspective to produce a new interpretation of modern human origins and our new results rooted the mtDNA tree in Asia (also Y). It is satisfying to know that we achieved this without bias from prior knowledge of your 1983 paper (I became aware of this much neglected paper through an Internet comment by a reader of our preprint Yuan et al 2017) or any non-molecular finds in Asia. And that is the way a molecular approach should be.

I enjoyed reading your review article in Cell, 2015. Needless to say, I like the last sentence at the end: “The unique features of the mtDNA may require a reassessment of some of our core assumptions about human genetics and evolutionary theory.” I respect you very much for being honest and indeed let’s work together to do exactly that. The present state of affairs is so very messy and confusing with so many ad hoc ‘epicycles’ that we are long overdue for a Copernican revolution (I am afraid that the second half of your letter did strike me as an epicycle for the mtDNA molecular clock). It is time for a scientist of your caliber to put words into action by taking the initiative and the honorable responsibility to organize a workshop. The goal would be to achieve a new level of consensus on assumptions and methodologies to benefit future research by comparing the present model with all other legitimate competing models. It may seem like self-serving but I am not aware of any other legitimate alternatives other than the Out of Asia model now independently discovered twice. We were really just rediscovering what you and your colleagues have found back in 1983 and we would not hesitate a bit to give you full credit for that. By the way, your 1983 paper was cited in the Chinese language literature just a handful of times in the past 34 years, and they were meant to promote the Out of Africa model with the Out of Asia message never touched.

I hope this helps you see our differences and which one makes more sense.

Best regards,

Shi