Friday, March 17, 2006

We're all Jesus' Children? Some more so than others probably   posted by the @ 3/17/2006 01:00:00 AM

Steve Olson of "Modelling the recent common ancestry of all living humans" fame has a piece in Slate discussing this finding with a lead-in related to the The Da Vinci Code trial:

On Monday Dan Brown, author of The Da Vinci Code, testified in a London courtroom to defend himself against the charge that he stole from an earlier book the idea that Jesus has a secret line of descendants who are alive today. But no matter how the court case turns out, both books are confused. If anyone living today is descended from Jesus, so are most of us on the planet.

He does a good job of explaining the highly reticulated mess that is true human genealogy. This is an important point and he deserves credit for the piece. Unfortunately, he fails to point out that even at the identical ancestor point, some people are more your ancestors than others. With that in mind, consider his closing remarks:

The risk of today's genetic genealogy tests is that they tend to divide people into groups, whereas the real message that emerges from genealogy is one of connections. For centuries, scientists have tried to sort people into biological categories. In the 18th and 19th centuries, they pounced on the idea of race and used it to formulate hypotheses about human differences that had disastrous social consequences. In the 20th century, scientists began to explore the greater complexities of our biological histories, which are impossible to capture in a word as simple-minded as "race." If genetic genealogy tests explored and explained these complexities, I'd have no problem with them. But most of today's tests hark back to the bad old days of racial science.

People may like to think that they're descended from some ancient group while other people are not. But human ancestry doesn't work that way, since we all share the same ancestors just a few millenniums ago. As that idea becomes more widely accepted, arguments over who's descended from Jesus won't result in lawsuits. And maybe, just maybe, people will have one less reason to feel animosity toward other branches of the human family.

Obviously it would be nice if the world were to move in that direction, but it's a bit of an overstatement. If Jesus is the ancestor of anyone living today, then most likely he's the ancestor of some people much more than others. While we do "all share the same ancestors just a few millenniums ago", this doesn't change the fact that some of us much more related to one another than others. In other words, most of the bite in "race" is still there.

Addendum from Razib: Two points to keep in mind, much of the human genome is redundant and does not exhibit sequence level polymorphism. Second, genes are discrete, that is, information will be lost via sampling, so the number of ancestors is less relevant than their proportional representation in the nodes of your genealogical tree. Olson in the Slate piece talks about comparing lists of ancestors and seeing that they have all the same people in them, but of course that's a lot less genetically relevant than the number of times they show up in the list.

Mortimer asks an interesting question about proving that the last common ancestor of all human beings lived a few thousand years ago. We "know" (via inferences from coalescent theory) that the last common ancestor of the nonrecombinant Y lived tens of thousands of years ago, and than the last common mtDNA lived a little earlier. Since these are male and female only direct lineages it is clear that simple a priori assumptions suggest that the last common ancestor via any sequence of male and female ancestors should be far more recent.

But how do you "prove" this? As I said, genetics is discrete, so just because someone is your ancestor doesn't mean that there is a identical-by-descent fragment within you from that given ancestor (a copy of unique sequence for example). I think the key is to look for a mutant which emerged recently, and if that mutant has fixed in the population, than by definition (let's exclude horizontal gene transfer) the line of ancestry has to go back to the individual in whom the mutation originated. Additionally, a wide survey of ancient remains might be able to get a general sense of the time frame when this mutation arose, when its frequence approaches statistically significant neglibility in the sample. The probability of fixation for a new allele within a population (a mutation) is about ~4Ne generations, where Ne is the effective breeding population for a neutral mutant. This is a gross simplification, and this is the expectation of the time to fixation, not a number etched in stone (there will be variance and some alleles fix earlier, some later), but the simplifying assumptions seem to suggest the time until fixation will be longer in geographically dispersed species than the prediction based on this equation. Even if Ne is assumed to be in the hundreds that seems too many generations to get the last common ancestor via sequencing on a neutral mutant.

So I think the best bet for smoking-gun-proof (as opposed to reasonable certainty based on a priori models) would be to look for positively selected alleles which have moved to fixation within the last 3,000 years. I don't know, or frankly think, that we will find anything like this across all extant populations. But that would be "proof," assuming we can find this super-selected allele, get a large sample of ancient remains, and are willing to accept a "good enough" answer which isn't absolutely definitive (I would bet that Tasmanian Aboriginals would be less likely to fix the magic allele because their small population implies that random genetic drift could swamp even large selection coefficients).

Anyway, I think this sort of "scientific genealogy" does tell us something about human psychology, if it doesn't tell us anything about genetics on a deep level. Consider this, we are conditioned to assume about 1/2 relationship to our siblings, 1/2 to each parent. Even in the case of 1st cousin marriage the parents are probably only 1/8 related to each other. Let's ignore cases of extreme inbreeding or clan cultures where the inbreeding coefficient is cranked up by generations of incest. Humans have certain expectations of their ancestors. You have two parents. Four grandparents. Eight great-grandparents. Do you know all your great-great-grandparents by heart? My understanding is that for most people there is little information in their heads about ancestors beyond 3 generations, though in many cultures a patrilineage or a "famous ancestor" might be noted. Our intuitions are conditioned by the last 3 generations, but as we move back up the family tree reticulation becomes much more of an issue. Our great-great-great-great-great-great-great-grandfather might also be our great-great-great-great-great-grandfather and our great-great-great-great-great-great-uncle, etc. We may have a fixed number of male and female ancestors, but some of those males and females are duplicates. Logically this is obvious, but I don't think people internalize this too deeply, otherwise there wouldn't be such emotional salience in regards to mtDNA or Y chromosomal tests which "prove" ancestry, though what they prove is rather trivial. And Olson has to be betting that people don't think about this much with his facile comparison of the particular ancestors we have all being in common. If someone tells you that two individuals have the same grandparents that gives you an immediate gestalt understanding of their relationship (or, if you tell them they share two grandparents). If someone tells you that individuals a & b share all the same ancestors that doesn't tell you much because the variance in ancestral contribution between the ancestors is far greater than would be between average grandparents (if siblings mated, you would have four grandparent slots occupied by two unique individuals, but we don't usually consider this case because it is so atypical).