Mitochondrial DNA analysis is limited

[politas]

There was some discussion recently about a study showing substantial differences between Neanderthal DNA and human DNA, and what that study actually means. The study in question examined mitochondrial DNA recovered from neanderthal remains and compared it to human mitochondrial DNA. It showed a branching from modern human mitochondrial DNA well before the rise of H. sapiens as a species.

Since this study only looked at mitochondrial DNA, it has serious limitations, and can only be considered as an indicator, not definite evidence of anything other than the specific conclusions stated. It does not prove that there was no mixing of neanderthal and sapiens genetic material

Here is a paper talking about the problems of trying to identify speciation histories and the limitations of mitochondrial DNA in such investigations. An important quote is the following:

...speciation histories based on mtDNA alone can be extensively misleading.

In the case of humans, for instance, we know that all modern humans can trace their mitochondrial DNA back to a single female from 150,000-250,000 years ago. This in no way means that there was only a single human female alive at that time. Nor does it mean that none of the other human females alive at that time have contributed their DNA to humans living today. It simply means that the line of direct female ancestry for everyone traces to that single individual.

It is entirely possible that if neanderthals and sapiens were genetically compatible, there is some neanderthal-originating DNA existing in modern human populations.

 

[Dave1001]

There was some discussion recently about a study showing substantial differences between Neanderthal DNA and human DNA, and what that study actually means. The study in question examined mitochondrial DNA recovered from neanderthal remains and compared it to human mitochondrial DNA. It showed a branching from modern human mitochondrial DNA well before the rise of H. sapiens as a species.

Since this study only looked at mitochondrial DNA, it has serious limitations, and can only be considered as an indicator, not definite evidence of anything other than the specific conclusions stated. It does not prove that there was no mixing of neanderthal and sapiens genetic material

Here is a paper talking about the problems of trying to identify speciation histories and the limitations of mitochondrial DNA in such investigations. An important quote is the following:

In the case of humans, for instance, we know that all modern humans can trace their mitochondrial DNA back to a single female from 150,000-250,000 years ago. This in no way means that there was only a single human female alive at that time. Nor does it mean that none of the other human females alive at that time have contributed their DNA to humans living today. It simply means that the line of direct female ancestry for everyone traces to that single individual.

It is entirely possible that if neanderthals and sapiens were genetically compatible, there is some neanderthal-originating DNA existing in modern human populations.

I still think that if we have no mDNA heritage from neanderthals, it implies very limited reproductive contact. Otherwise, counterintuitively, it would mean that no one on the direct maternal line of any neanderthals survived to modern times, but yet neanderthals genes contribute in a more than very limited way to our current genetic make up. How would that happen in practice? Modern human women preferred neanderthal men, but modern human men didn't find neanderthal women attractive? Sons of neanderthal women and modern human men reproduced more successfully than daughters of neanderthal women and modern human men?

Unrelatedly, I would think that modern woman's mDNA would probably have been pretty robust relative to her peers since she's the direct maternal line of every humanoid living on the Earth today, from a time period of when there were at least thousands of humanoid females. Sort of like how everyone 150,000 years from now will probably have Barbara Bush as a direct maternal ancestor.

 

[brooklyn44]

David

Sort of like how everyone 150,000 years from now will probably have Barbara Bush as a direct maternal ancestor.

bite your tongue

 

[brooklyn44]

oops i lost my quote box...sorry

 

[SkepticScott]

As Richard Dawkins pointed out in his "The Ancestor's Tale", if you pick a person back far enough, either all living people will be descended from that person, or none of them will.

 

[politas]

As Richard Dawkins pointed out in his "The Ancestor's Tale", if you pick a person back far enough, either all living people will be descended from that person, or none of them will.

And at any particular point from that point and further back, there will be thousands of people who are ancestors of everybody.

At the time of Mitochondrial Eve, there were thousands of humans alive. Mitochondrial Eve is our common ancestor for all mtDNA, but she is not our common ancestor for all nDNA. Thousands of other human females living at the same time as Mitochondrial Eve are also ancestors of every living human, even though we have no mtDNA from them.

Mitochondrial DNA is not a representative model for nDNA.

 

[Dave1001]

And at any particular point from that point and further back, there will be thousands of people who are ancestors of everybody.

At the time of Mitochondrial Eve, there were thousands of humans alive. Mitochondrial Eve is our common ancestor for all mtDNA, but she is not our common ancestor for all nDNA. Thousands of other human females living at the same time as Mitochondrial Eve are also ancestors of every living human, even though we have no mtDNA from them.

Mitochondrial DNA is not a representative model for nDNA.

Yes, but would it be expected that all living hominids today share the mtDNA of only one hominid from 150K years ago? Or is it statistically improbable? Also please explain why either way?

 

[politas]

Yes, but would it be expected that all living hominids today share the mtDNA of only one hominid from 150K years ago? Or is it statistically improbable? Also please explain why either way?

I'm not sure what you're asking. It is not a statistical probability that all humans have mtDNA descended from (not exactly the same) a single hominid from 150K years ago. It's a fact. Ie, it has been proven by scientific investigation beyond any reasonable argument.

The probability of a single mtDNA ancestor being found is 1. The only question is exactly when that ancestor will be found, and calculating those probabilities has far more to do with geographic separation of human populations than anything else. How is that relevant?

What is relevant is that nuclear DNA descends through both male and female lines (except for certain parts of the Y-chromosome), so individuals which are cut out of mtDNA descent (due to having only male offspring) are not excluded from nDNA descent.

 

[Dave1001]

I'm not sure what you're asking. It is not a statistical probability that all humans have mtDNA descended from (not exactly the same) a single hominid from 150K years ago. It's a fact. Ie, it has been proven by scientific investigation beyond any reasonable argument.

The probability of a single mtDNA ancestor being found is 1. The only question is exactly when that ancestor will be found, and calculating those probabilities has far more to do with geographic separation of human populations than anything else. How is that relevant?

What is relevant is that nuclear DNA descends through both male and female lines (except for certain parts of the Y-chromosome), so individuals which are cut out of mtDNA descent (due to having only male offspring) are not excluded from nDNA descent.

Let me rephrase the question. Let's say you start out with the number of female hominids that existed on Earth 150,000 years ago. Each has mtDNA that is distinguishable from the otherr. What are the odds, all things being equal, that all hominids 150,000 later would have the mtDNA of the same one female that was living at that time? Of the top of my head, I'd think that the basic odds would be (1) : (the number of female hominids alive 150,000 years ago squared). In other words, if there were 1,000 or more female hominids alive 150,000 years ago, which I think there likely were at minimum, the odds that today we would all share the mtDNA of one of them and only one of them is 1 in a million or lower. I know that factually we do, I'm discussing the apparent statistical improbability of this outcome. Am I missing something here? If so, please explain in detail.

 

[Cuddles]

Of course, all arguments about "Mitochondrial Eve" are based on the assumption that mitochondrial DNA is only passed down the female line. Since that now appears not to be the case the whole argument is moot.

http://www.newscientist.com/article/dn2716.html
http://content.nejm.org/cgi/content...0&volume=347&firstpage=576&resourcetype=HWCIT

 

[Dave1001]

Of course, all arguments about "Mitochondrial Eve" are based on the assumption that mitochondrial DNA is only passed down the female line. Since that now appears not to be the case the whole argument is moot.

http://www.newscientist.com/article/dn2716.html
http://content.nejm.org/cgi/content...0&volume=347&firstpage=576&resourcetype=HWCIT

I haven't had a chance to peruse those links yet, but if we can get mtDNA from either males or females, and all of us have mtDNA fully traced to a single ancestor 150,000 years ago, I think that would make that outcome even more statistically improbable? Yes or no, and why?

(Sidenote: I already understand mathematically why if you go back far enough in time, for any ancient individual either we all have them as an ancestor, or none of us do, although I think that's technically not inevitable because there theoritcally could have been some completely endogamous subpopulations of humanity for at least the past few tens of thousands of years.)

 

[Cuddles]

I haven't had a chance to peruse those links yet, but if we can get mtDNA from either males or females, and all of us have mtDNA fully traced to a single ancestor 150,000 years ago, I think that would make that outcome even more statistically improbable? Yes or no, and why?

The results are interpreted assuming that all mtDNA is from the female line. If this assumption is false then all analysis is wrong. I haven't seen any of the raw data and I don't know how a new analysis would work without this assumption so I can't really say if the outcome would be more improbable. I would think that such an analysis would be virtually impossible since instead of a single continuous female line there would be a lot of mixing which would obscure any such line.

 

[politas]

Let me rephrase the question. Let's say you start out with the number of female hominids that existed on Earth 150,000 years ago. Each has mtDNA that is distinguishable from the otherr. What are the odds, all things being equal, that all hominids 150,000 later would have the mtDNA of the same one female that was living at that time? Of the top of my head, I'd think that the basic odds would be (1) : (the number of female hominids alive 150,000 years ago squared). In other words, if there were 1,000 or more female hominids alive 150,000 years ago, which I think there likely were at minimum, the odds that today we would all share the mtDNA of one of them and only one of them is 1 in a million or lower. I know that factually we do, I'm discussing the apparent statistical improbability of this outcome. Am I missing something here? If so, please explain in detail.

No, the chance is way more probable than that. You're looking at it from the wrong perspective.

I believe that 150k years ago is far enough back that the population is divided into those who are ancestors to everybody, and those who are ancestors to nobody. Therefore:

For the existing population at 150k years ago, we therefore start with the female population which are ancestors. This number must be less than the number of living females at the time. Examine the case of each female ancestor. Each generation linking this ancestor to people living today can be either male or female. If any single generation consists of all males, that female cannot be a mtDNA source (or at least, it is highly unlikely, given the rare inclusions of paternal mtDNA). For each female therefore, the chance of being an mtDNA ancestor is quite small, since a 50% chance of a male descendant must be multiplied by the number of descendants in each generation and those chances added to calculate the chance of missing out on being an mtDNA ancestor. (in fact, the calculation is a little more complicated, since spreads across different generations can result in breaking the mtDNA descent chain without a single generation needing to be all male.) As soon as you start adding up a large number of small probabilities, you end up with one large probability. Let's say the average chance of an all-male generation is one in a thousand. Since the number of generations is on the order of 6,000, that would make the probability 6, ie a certainty. Obviously, that's a completely bogus calculation using numbers I plucked out of thin air, but it shows the way the probabilities trend.

Clearly, at least one female must be an mtDNA ancestor, but the chance of any particular female being an mtDNA ancestor is smaller the further back in time you go. So as the distance in time increases, the probability of having a single mtDNA ancestor approaches 1.

Does that clear things up?

 

[politas]

Of course, all arguments about "Mitochondrial Eve" are based on the assumption that mitochondrial DNA is only passed down the female line. Since that now appears not to be the case the whole argument is moot.

http://www.newscientist.com/article/dn2716.html
http://content.nejm.org/cgi/content...0&volume=347&firstpage=576&resourcetype=HWCIT

That article states that they expect paternal DNA inheritance to be very rare, which would really only adjust the probabilities slightly. Say changing the probability of mtDNA descent through a particular ancestor from .5 to .499 if paternal mtDNA inheritance happens once in every thousand cases.

That's not going to have a huge impact, though it will certainly have some. I can't predict exactly what will come of it, though, so my argument may be invalidated. I don't think the current information does that, though.

 

[casebro]

So, your father's mtDNA comes from his mother, who WAS a decendent of Eve... vs coming from your MOTHER, who was a decendent of Eve...

 

[RSLancastr]

I certainly hope that funding for study of mitochondrias increases, because, as I understand it, they are what allows a Jedi to control the Force.

 

[Dave1001]

No, the chance is way more probable than that. You're looking at it from the wrong perspective.

I believe that 150k years ago is far enough back that the population is divided into those who are ancestors to everybody, and those who are ancestors to nobody. Therefore:

For the existing population at 150k years ago, we therefore start with the female population which are ancestors. This number must be less than the number of living females at the time. Examine the case of each female ancestor. Each generation linking this ancestor to people living today can be either male or female. If any single generation consists of all males, that female cannot be a mtDNA source (or at least, it is highly unlikely, given the rare inclusions of paternal mtDNA). For each female therefore, the chance of being an mtDNA ancestor is quite small, since a 50% chance of a male descendant must be multiplied by the number of descendants in each generation and those chances added to calculate the chance of missing out on being an mtDNA ancestor. (in fact, the calculation is a little more complicated, since spreads across different generations can result in breaking the mtDNA descent chain without a single generation needing to be all male.) As soon as you start adding up a large number of small probabilities, you end up with one large probability. Let's say the average chance of an all-male generation is one in a thousand. Since the number of generations is on the order of 6,000, that would make the probability 6, ie a certainty. Obviously, that's a completely bogus calculation using numbers I plucked out of thin air, but it shows the way the probabilities trend.

Clearly, at least one female must be an mtDNA ancestor, but the chance of any particular female being an mtDNA ancestor is smaller the further back in time you go. So as the distance in time increases, the probability of having a single mtDNA ancestor approaches 1.

Does that clear things up?

Great explanation. I didn't read it with a fine tooth comb but it makes sense in principle. So, why don't all men share a Y chromosome from a single male hominid also about 150K years ago? My understanding is that if we all share a Y chromosome from a single male hominid, he's from a different era tens of thousands of difference in time than "Eve". If so, do those time discrepencies correllate with what we'd expect from the lower odds of having boys than girls? And other aspects of sexual behavior?

 

[politas]

Great explanation. I didn't read it with a fine tooth comb but it makes sense in principle. So, why don't all men share a Y chromosome from a single male hominid also about 150K years ago? My understanding is that if we all share a Y chromosome from a single male hominid, he's from a different era tens of thousands of difference in time than "Eve". If so, do those time discrepencies correllate with what we'd expect from the lower odds of having boys than girls? And other aspects of sexual behavior?

Well, as I said, I have no idea what the exact probabilities are, but the probability of a single mtDNA (or Y-chomosome nDNA) ancestor certainly increases as you go back in time. I think the Y-chomosome "Adam" is something like a hundred thousand years earlier than mitochondrial Eve. (I can check on that if you want, but it really isn't all that important. The dating of Mitochondrial Eve is actualy a pretty vague figure inside a 100K year range) Probabilities are funny things. Somteimes less probable things happen. There's no particular reason for Mitochondrial Eve and Y-chromosome Adam to be contemporaries.

Now do you see why the study of mtDNA in neanderthals cannot really say all that much about intrusion of neanderthal nDNA in H. Sapiens?

 

[Dave1001]

Well, as I said, I have no idea what the exact probabilities are, but the probability of a single mtDNA (or Y-chomosome nDNA) ancestor certainly increases as you go back in time. I think the Y-chomosome "Adam" is something like a hundred thousand years earlier than mitochondrial Eve. (I can check on that if you want, but it really isn't all that important. The dating of Mitochondrial Eve is actualy a pretty vague figure inside a 100K year range) Probabilities are funny things. Somteimes less probable things happen. There's no particular reason for Mitochondrial Eve and Y-chromosome Adam to be contemporaries.

Now do you see why the study of mtDNA in neanderthals cannot really say all that much about intrusion of neanderthal nDNA in H. Sapiens?

Probabilities are good indicators. I think pretty much all of empiricism is based on this as an underlying principle right? I suppose with hominids we only have one example, but in principle this should be expandable to most other species, right? I would think if we knew how long the generation span was of a species was for a dimorphic sperm/egg sex species, we could tell about how many generations back on its maternal sides to its most likely "Eve" and how many generations back on its paternal sides to its most likely "Adam", and then we could verify these predictions empirically, at least for those species that also get almost all their mtDNA maternally and have Y chromosomes that they get paternally. I would think for most such species the predictions should match the observed results. What do you or others think about that?

By the way, I haven't been asking this strain of questions to try to get an answer on the neanderthal nDNA post. This is all just knowledge for the sake of knowledge, so don't feel obligated to tie it back in to that: I understood your point the first time you posted it on that side topic.

 

[politas]

Probabilities are good indicators. I think pretty much all of empiricism is based on this as an underlying principle right? I suppose with hominids we only have one example, but in principle this should be expandable to most other species, right? I would think if we knew how long the generation span was of a species was for a dimorphic sperm/egg sex species, we could tell about how many generations back on its maternal sides to its most likely "Eve" and how many generations back on its paternal sides to its most likely "Adam", and then we could verify these predictions empirically, at least for those species that also get almost all their mtDNA maternally and have Y chromosomes that they get paternally. I would think for most such species the predictions should match the observed results. What do you or others think about that?

Yes, most empiric scientific facts are when examined closely, no more than a probability. A probability which is beyond reasonable doubt.

This is the basis of most "Intelligent Design" claims, they feel that certain claims of Neo-Darwinistic evolution are too improbable for reasonable belief. In most cases, their arguments fail for similar reasons as your misunderstanding - faulty statistical analysis. Statistical analysis and probabilities often appear relatively simple and obvious while hiding levels of complexity which can entirely reverse the expected trends.

In the case of mtDNA / Y-chromosome analysis, calculating probabilities and expected timing of common ancestors is immensely complex. Factors such as spreads of separate populations, interaction boundaries between populations, mutation rates, longevity/generation rate,...

I wouldn't know how to start in calculating actually when a common mtDNA ancestor would be expected for any particular species. I'm not sure that the effort is even worthwhile, since there would almost have to be a lot of assumptions included in the calculation, quite possibly enough to make the estimation unfounded.

Discovering actually when a particular species' common mtDNA ancestor existed is actually a lot easier than calculating the probabilities, since that calculation is entirely based on actual observed data.