Tuesday, September 25, 2012

Human Mutation Rate Is Gender Biased And Low

A new study published this month at Nature Genetics confirms with a direct measurement, multiple lines of evidence suggesting that mutation rates in modern humans are about half what had been assumed based on a presumed human-chimpanzee divergence date by most of the last decade's genetics papers. This factor of two difference is sufficient to make almost every interesting conclusion about pre-history that could be drawn from genetic data using the old dates wrong.

More intriguingly, it also suggests that about 85% of new mutations come from fathers, rather than mothers. While it has long been known that advanced paternal age is associated with new disease causing mutations, and the likely culprits behind this discrepancy have been identified, the general bias towards mutations having a paternal rather than a maternal source has been less widely appreciated.

The key line of the paper states: "we obtained an SNV mutation rate of 1.20 × 10−8 (95% confidence interval 0.89–1.43 × 10−8) mutations per base pair per generation. . . . there was strong evidence (P = 2.67 × 10−4) for a paternal bias in the origin of new mutations (85% paternal)."

For most of the genome, this gender bias isn't very important when trying to determine the age of a mutation. Everyone has a mother and a father. Everyone gets about half of their autosomal genome from each parent. So, gender differences in mutation rates all average out.

But, this may indicate that the proper mutation rate to use when determining the actual historic date when a mutation arose based on mutations observed in non-recombining Y-DNA may be higher than it is when attempting to estimate a historic date based upon mutations in autosomal DNA. Likewise, the mutation rates for autosomal X chromosomes may be somewhat lower than for autosomal DNA generally, because a disproportionate share of its lineage, although not 100% as in mtDNA, is from mother to child transmissions that are less mutation prone.

The difference in mutation rates in father to child genetic transmission as opposed to mother to daughter transmission could also adds to our understanding of the apparent discrepancies between the age of the root of the Y-DNA tree and estimates of the historic era of the most recent common ancestor of modern humans based on other population genetic methods (although this seems to make the problem worse rather than better).

11 comments:

terryt said...

"More intriguingly, it also suggests that about 85% of new mutations come from fathers, rather than mothers".

As you say, that has long been known, though perhaps not widely appreciated. The general explanation is that sperm is manufactured all through a male's life whereas eggs have basically been completed at the birth of a female. Hence more opportunity for mutation in the male line.

"The difference in mutation rates in father to child genetic transmission as opposed to mother to daughter transmission could also adds to our understanding of the apparent discrepancies between the age of the root of the Y-DNA tree and estimates of the historic era of the most recent common ancestor of modern humans based on other population genetic methods"

In fact there is no need at all to assume the ancestral modern mtDNA originated in the same place as the ancestral Y-DNA. Or at anywhere near the same time. A population of humans didn't suddenly become 'modern' over night. The process took some time. In fact we actually can detect a slightly different region of origin for the basal mtDNA and Y-DNA lines. Modern Y-DNA looks most likely to have first appeared in Cameroon evidently, while the case for modern mtDNA's origin can be most easily made for somewhere further east, perhaps at least the Bahr el Ghazal. Perhaps the origin of 'modern' humans lies in the mixing of two populations in the first place. Hybrid vigour?

andrew said...

"there is no need at all to assume the ancestral modern mtDNA originated in the same place as the ancestral Y-DNA. Or at anywhere near the same time."

It doesn't need to be the same place or the same time (and surely isn't if you are fine grained enough), but haplogroups that persist to the present of Y-DNA have to derive from a coherent community, haplogroups that persist to the present of mtDNA have to derive from a coherent community, and the levels of correlation in modern populations between Y-DNA haplogroups and mtDNA haplogroups in similar phylogenetic positions, as well as the kind of extreme demographic secnarios necessary to produce the complete selective/replacement sweeps in patrilines but not matrilines or visa versa in cases where there is no other population genetic indication of those kind of demographic scenarios all favor a more parsimonious scenario in which Y-DNA haplogroups and mtDNA haplogroups are emerging in parallel within thousands of years of each other within the same communities which are populations that start out relatively compact geographically. Some discrepencies make sense. Roots of the tree separated by factors of two that involve periods of time that may exceed 100,000 years aren't a good fit to the genetic and non-genetic data taken as a whole. The only scenarios in which that big of a discrepency makes much sense based on the kind of gradual modernity and hybrid vigor kinds of scenarios you suggest are those with much smaller effective population sizes for the first half of modern human population genetic history than are commonly assumed - perhaps in the low four digits, rather than the high four digits or low five digits with no population bottlenecks in Africa as most studies have suggested.

terryt said...

"but haplogroups that persist to the present of Y-DNA have to derive from a coherent community"

But the makeup of that community can be continually changing. That seems actually to be the point Dienekes is currently trying to make. People move about.

"a more parsimonious scenario in which Y-DNA haplogroups and mtDNA haplogroups are emerging in parallel within thousands of years of each other within the same communities which are populations that start out relatively compact geographically".

But the association of particular haplogroups keeps changing. For example Y-DNA F seems closely associated with mt-DNA M as they both moved through South Asia, but F-derived Y-DNA MNOPS seems then to have expanded with N-derived mt-DNA R. The haplogroups swapped. Y-DNA MNOPS and mt-DNA R may have first started out 'relatively compact geographically', but there is no reason to assume so. Members of Y-DNA MNOPS may have moved independently into a region where members of mt-DNA R were already present. So the region from where mt-DNA R and Y-DNA MNOPS expanded may not have been all that compact geographically. Quite spread out in fact. And we know that in any particular region Y-DNA usually shows evidence of more recent ancestry that mt-DNA.

"Some discrepencies make sense. Roots of the tree separated by factors of two that involve periods of time that may exceed 100,000 years aren't a good fit to the genetic and non-genetic data taken as a whole".

Not a 'good' fit, but not an 'impossible' one.

"The only scenarios in which that big of a discrepency makes much sense based on the kind of gradual modernity and hybrid vigor kinds of scenarios you suggest are those with much smaller effective population sizes for the first half of modern human population genetic history than are commonly assumed"

Not necessarily so. Just subsets of any population would usually be involved in any hybridism. That would considerably reduce the effective population sizes. But from that small hybrid population genes could spread out through the species in almost any direction, leading to the gradual move towards 'modernity' that we actually see.

terryt said...

Interesting post on inbreeding (via John Hawks):

http://io9.com/5863666/why-inbreeding-really-isnt-as-bad-as-you-think-it-is

"In point of fact, we're all technically inbred, if you go back far enough, because simple math demands that we have to be. Our number of ancestors grows exponentially with each generation, from two parents to four grandparents to eight great-grandparents, and so on. In less than a thousand years, you've accumulated tens of billions of ancestors, more than the amount of humans who have ever lived on this planet".

That inbreeding is why we have species and not some amorphous series of clines.

"Professor Alan Bittles, an adjunct professor at the Centre for Comparative Genomics at Australia's Murdoch University, who has worked on the subject for over three decades and in 2008 conducted a review of forty-eight studies from eleven countries on the rate of birth defects in the children of first cousins".

Obviously cousins who share nothing more than just one of a pair of siblings as parents are not really very 'inbred' at all though. It's when the same individual shows up on both sides of a pedigree just one or two generations back that we can really talk about inbreeding.

"But still, while we all carry the genes for such potentially deadly conditions, not all autosomal recessive disorders are so easily activated, with many requiring multiple generations of inbreeding before becoming a serious problem. There does tend to be a gradual decrease in reproductive fitness and general health - children of inbreeding tend to have more trouble having kids and are slightly sicklier, and that gets worse over time"

But in the case of the Paleolithic we are talking very long periods of time and relatively small populations. Inbreeding would have become a problem in many regions.

"That repeated division means that by the fifteenth generation - which is only a few centuries ago - your average ancestor (assuming zero inbreeding) is contributing, on average, less than a single gene to your current genome. Go back a thousand years to the 30th generation, and the average genetic contribution is effectively zero".

I offered my take on this last subject some years back:

http://humanevolutionontrial.blogspot.co.nz/2009/06/human-evolution-on-trial-pedigrees.html

And on the subject of inbreeding:

http://humanevolutionontrial.blogspot.co.nz/2009/06/human-evolution-on-trial-hybrid-vigour.html

terryt said...

Apologies for carrying on, but there are a few things I feel you've overlooked.

"the kind of extreme demographic secnarios necessary to produce the complete selective/replacement sweeps in patrilines but not matrilines or visa versa in cases where there is no other population genetic indication of those kind of demographic scenarios all favor a more parsimonious scenario in which Y-DNA haplogroups and mtDNA haplogroups are emerging in parallel"

But we can be absolutely sure that such selective sweeps have happened. For a start there were obviously haplogroups in Eurasia other than 'modern' ones when thee modern ones first emerged from Africa. These other haplogroups have now been completely replaced, selectively. And actually any of the basal modern lines could have been sustained within such populations after they had left Africa. So the mt-DNA and Y-DNA lines can be remarkably independent.

"the first half of modern human population genetic history than are commonly assumed ... the high four digits or low five digits with no population bottlenecks in Africa as most studies have suggested".

Whatever the size of the 'original' population, whether in the high four digits or low five digits, it must also have suffered complete selective/replacement sweeps in both patrilines and matrilines. Not everyone in any such an 'original' population would have had the same haplogroups. It is extremely unlikely that some small group of humans woke up one morning to find they all had the same mt-DNA and the men all had the same Y-DNA. Other haplogroups must have been present in that hypothetical population. Those haplogroups have since been eliminated. Neighbours presumably shared at least some of those non-modern haplogroups, so presumably the 'original' population was quite capable of breeding with their near neighbours. In fact we can now be fairly sure they were able to interbreed with very distant neighbours, even some from outside Africa. I agree wholeheartedly with Dienekes recent comment:

"I am more or less convinced that admixture between very divergent populations of Homo heidelbergensis played a major role in shaping modern humans".

Andrew Oh-Willeke said...

"there were obviously haplogroups in Eurasia other than 'modern' ones when thee modern ones first emerged from Africa."

Not at all necessarily so. Indeed, migration by very uniparental genetically homogenous populations by tribes or bands of people made up of small numbers of extended families that are deeply interrelated to each other in a demographic arrangement which, for want of a more familiar example, resembles the twelve tribes of Israel in Exodus, seem to be the more common pattern.

Migrations involving a meaningfully mixed hg distribution seem to be very rare prior to the Bronze Age or later.

I also saw a paper this week exploring the possibility of modern human population size staying very low (in the low single digit thousands) pretty much up until Out of Africa. So maybe those demographics are present.

With realistic demographic assumptions in modern human genetic history it is very hard to purge a selectively neutral uniparental genetic marker that is present at more than low single digit percentages from a population. Haplogroups tend to reach fixation rapidly and then drift in percentage only very mildly over long periods of time. For example, we still have discernable mtDNA/Y-DNA traces of admixture at the Paleolithic/Neolithic boundary in Europe.

Strong founder effects associated with migrations by extended family tribes are a much more plausible mechanism for purifying mtDNA and Y-DNA mixes than selective/replacement sweeps.

And there is no good reason to think that any of the ancestry informative Y-DNA or mtDNA haplogroups of members of a population genetic community that has reached fixation of haplogroup frequency has any selective effect whatsoever.

terryt said...

"Not at all necessarily so [haplogroups in Eurasia other than 'modern' ones]".

Do you mean to say that Neanderthals and H. heidelbergensis did not have Y-DNA or mt-DNA haplogroups? Can't see how I'd agree with that. Obviously those non-modern haplogroups have been replaced by a selective sweep.

"migration by very uniparental genetically homogenous populations by tribes or bands of people made up of small numbers of extended families that are deeply interrelated to each other in a demographic arrangement which, for want of a more familiar example, resembles the twelve tribes of Israel in Exodus, seem to be the more common pattern".

True, but the haplogroups contained within such migrations have often replaced previous ones. Even 'modern' ones in some cases. The members of such migrations have usually ultimately settled down and mixed with the previous inhabitants however. Later expansions of 'tribes' from the region have then often contained a mix of the 'original' and the 'recent' haplogroups.

"Migrations involving a meaningfully mixed hg distribution seem to be very rare prior to the Bronze Age or later".

Perhaps rare, but by no means unknown. The Polynesians are made up of Y-DNA from Southern Wallacea and mt-DNA from further north, presumably having been originally members of separate populations. The expansion of Y-DNA MNOPS from SE Asia too seems to be associated with a change in accompanying mt-DNA.

"With realistic demographic assumptions in modern human genetic history it is very hard to purge a selectively neutral uniparental genetic marker that is present at more than low single digit percentages from a population".

To me the evidence is overwhelming that human haplogroups are not 'selectively neutral'. They are selected for, not through genetics, but through the expansion of technologies or cultures. You hint at agreement with that position with your statement, 'Migrations involving a meaningfully mixed hg distribution seem to be very rare prior to the Bronze Age or later'. Surely such replacement was not 'genetic' but 'technological'. There is no reason not to suspect that earlier replacements were the result of other than similarly technological expansions.

"For example, we still have discernable mtDNA/Y-DNA traces of admixture at the Paleolithic/Neolithic boundary in Europe".

And we still have discernible traces of aDNA from Neanderthals. So we must conclude that Neanderthal haplogroups have been replaced.

"Strong founder effects associated with migrations by extended family tribes are a much more plausible mechanism for purifying mtDNA and Y-DNA mixes than selective/replacement sweeps".

Haplogroup 'selective/replacement sweeps' in humans is presumably usually the product of 'strong founder effects associated with migrations by extended family tribes'.

"And there is no good reason to think that any of the ancestry informative Y-DNA or mtDNA haplogroups of members of a population genetic community that has reached fixation of haplogroup frequency has any selective effect whatsoever".

I would say that the evidence points strongly to such having happened in South China and SE Asia. Y-DNA K had reached fixation in the region before O expanded, and largely replaced it. K (and C) survived in regions that had been already well populated though, but the haplogroup(s) remains today through the region as very much a minority haplogroup. That replacement was through expansion of technology, in the form of the early Chinese Neolithic.

terryt said...

Sorry, a bit more:

"I also saw a paper this week exploring the possibility of modern human population size staying very low (in the low single digit thousands) pretty much up until Out of Africa".

Were the scientists involved influenced by Genesis, and the idea that all species derive from a single pair on an ark? I feel it is very difficult for even scientists to look at the evidence other than through the eyes they developed during their childhood.

The idea species evolve suddenly from small, genetically isolated populations owes everything to the mythical beliefs that lie at the base of religious belief. Such a belief cannot stand close inspection. Even the 'origin of life' involves hybridism. The basal lines of Archaea, Bacteria and Eucaryotes swapped genes at their origin. Presumably they were closely related at that time. The three groups do not descend from just small, genetically isolated populations. There is also research that suggests chimpanzees and and what gave rise to the human line swapped genes several times during their early separation. The Genesis theory of species origin is well and truly dead.

andrew said...

"Were the scientists involved influenced by Genesis, and the idea that all species derive from a single pair on an ark?"

Very unlikely.

The population genetic model that produces the mainstream estimate in the five figures is extremely simplified, has never been rigorously calibrated to see if it matches empirical data for megafauna of any kind (including humans), and is long overdue for reconsideration. Even if the conclusion of the old estimates is correct and the new paper is wrong, it is because the considerations ignored by the vastly oversimplified method of making the estimate happen to cancel out, not because the model itself really accounts substantively for all the moving parts that could impact ancient effective population sizes.

The notion of everyone deriving from a common ancestor of a particular gender at some point is a natural conclusion of tracing uniparental lineages back to a most recent common ancestor. The notion that a most recent common ancestor of patriline ancestor and a most recent common matriline ancestor of women might reasonably close in time doesn't require religious belief.

Extending ideas about the origin of life which involved organisms with asexual reproduction modes (something that even modern sharks and snakes continue to have) isn't very relevant to species like hominins that lack that mode of reproduction where horizontal transfer of genes via germ line bacteria doesn't appear to have played a particularly important role in the last 250,000 years or so in modern human evolution.

Polynesian migration mostly happened during or after the Bronze Age in Europe, and like the European Bronze Age involved a late Neolithic population that had made major technological advances over early Neolithic populations in the area.

Re Inbreeding: The optimal level of inbreeding for fitness, empirically, appears to be third cousin marriage, with fitness declining a bit among less closely related couples. Also, inbreeding can also lead (through a most unpleasant intermediate process) to genetic load dumping. A population that is inbred for a long time loses most of its harmful recessive genes eventually after enough generations, in the same way that harmful dominant genes typically disappear in the generation that they appear or perhaps one generation later.

andrew said...

The fact that one population which ancestrally informative genetic markers replaces another species through a selective sweep doesn't mean that those particular ancestrally informative genetic markers themselves are anything but selectively neutral. The distinction between a marker itself providing selective advantage and a marker being ancestry informative of participation in a population that has some sort of selective advantage at least at the group level is critically important for evaluating what current hg mixture proportions say about past admixture and introgression rates. If the ancestry informative marker itself is fitness enhancing, then it will come to be predominant regardless of the hg that was predominant at the time of admixture. A population that is 1% substrate and 99% superstrate, and one that is 99% substrate and 1% superstate will look identical if the AIM is not selectively neutral in the population twenty generations level without regard to whether the advantageous hg is in the superstrate or the substrate population. In contrast, if the AIM is selectively neutral, the hg mix after twenty generations will look very similar to the hg mix after three to five generations and will be highly influenced by the initial admixture proportions.

I've never argued that replacement or near replacement of one group by another doesn't happen, surely it does.

And I have devoted quite a few posts to discussing what kind of admixture models could leave significant Neanderthal and Denisovan admixture without any Neanderthal and Denisovan Y-DNA or mtDNA. A key point in those models is that the archaic uniparental markers are already complete absent by the second generation and that random genetic drift that purges these uniparental markers over time is not the relevant mechanism.

My point is that selective sweeps/replacement operates pretty much exclusive at the level of one migrating group replacing some other previously resident group. It does not happen very often at all within a particular group that persists over time integrating outsiders gradually. One uniparental hg within a population mix usually does not get purged out by random drift within that continous population over time or because one Y-DNA or mtDNA hg has a fitness enhancing property relative to another. The uniparental hg change happens because one tribe goes extinct or nearly so (often at the hands of their replacements) and a new tribe moves in.

terryt said...

" The notion that a most recent common ancestor of patriline ancestor and a most recent common matriline ancestor of women might reasonably close in time doesn't require religious belief".

But it is not necessarily true. Modern humans seem to have developed from the interaction of at least several populations. For example the MRCA of Y-DNA looks to have been somewhere near Cameroon yet the MRCA mt_DNA looks to have been further east in Africa. In which case we have hybrid vigour right from the very origin of modern humans.

" A population that is inbred for a long time loses most of its harmful recessive genes eventually after enough generations"

Only true if it actually survives which, in many cases, it does not.

" The distinction between a marker itself providing selective advantage and a marker being ancestry informative of participation in a population that has some sort of selective advantage at least at the group level is critically important for evaluating what current hg mixture proportions say about past admixture and introgression rates".

Very true. In fact I'd be reasonably sure that the mt-DNA and Y-DNA lines in modern humans have no genetic selection advantage at all. The selective advantage was in the technology and culture those lines carried.